This page contains discussions that have been archived from Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either start a new thread or use the talk page associated with that topic.
On two computers and multiple browsers, I'm suddenly seeing the title coordinates display much lower down than before. They nearly overlap infoboxes and the top line of text. Abductive (reasoning)19:27, 20 May 2021 (UTC)
This (and the above font-size problem) appears to be a direct result of the removal of the "mw-body-content" class from the "bodyContent" block and its addition to the "mw-content-text" block instead. Is this a MediaWiki issue or should a "fix" be applied (at least temporarily) to the local CSS? --Paul_012 (talk) 04:25, 21 May 2021 (UTC)
I'm wondering if it has anything to do with the upcoming software changes for Vector. A local fix would be great though because it's annoying. Buffaboytalk06:37, 21 May 2021 (UTC)
So, a minor update:
The issue impacting old Vector is indeed phab:T283206, the same as the above. I'm going to punt on updating anything regarding live coordinates since it looks like they are going to try to fix this soonly.
The issue impacting new Vector is the phab task I posted in this section.
It looks like this can indeed be fixed in new Vector also and safely relative to old Vector, I'll just need to learn how relative/absolute styling works.
{{coord}} when used with the parameter display=title hitherto placed coordinates for the article on the title line. Now coords are being displayed slightly below the title line, and are overwriting the top of infoboxes - example at Inchinnan Castle. Seems poor. The template itself has not been edited in recent times, so presumably the change is somewhere else. --Tagishsimon (talk) 09:54, 22 May 2021 (UTC)
LaundryPizza03, when categories on a template are edited, the changes don't aren't immediately applied to pages that use it. You'll either have to wait until the changes are processed or perform a WP:NULLEDIT, but since there's no reason to immediately apply the changes, you should just wait. – BrandonXLF (talk)01:33, 25 May 2021 (UTC)
Attempting to post image on Wikipedia, username is not linked to Wikimedia Commons
Hi everyone!
I am trying to upload some images onto a stub, but when I try to upload them (whilst logged into Wikipedia), my username shows up in red and does not link to my User page.
For example:
Username error
Thanks :)
{{information
|Description={{en|1= Screenshot of username error. Chrome 90.0.4430.212 on Mac}}
|Source=Screenshot
|Date=2021-05-25
|Author=Wikipedia Authors, see the history of [[w: Rice production in China]]
|Permission={{Wikipedia-screenshot|1=en|logo=no}}
}}
Commons is a different website (a separate instance of the MediaWiki software), even though accounts are unified among them. You have a separate User: page there vs here. Wikilinks on any given page default to linking to the pagename at the same site. If you want to point to the "other" site, you would have to encode that specifically in the link. See for example [1]. DMacks (talk) 03:55, 25 May 2021 (UTC)
commons:User:Avoandtoast was created by the alt account Avoandtoast1. "One user creating another user's userpage", and especially if variants on the name from newish editors, are common (sorry:) patterns of non-constructive contributions and could reasonably go by U1 there. Contesting that deletion "that other account is mine also, I was trying to diagnose a linking problem" or simply creating it using the same account should solve that concern. DMacks (talk) 04:10, 25 May 2021 (UTC)
Avoandtoast, I see you created your global user page and your username has turned blue, so that's resolved. You can log in on most Wikimedia projects (like Wikipedia, Wikimedia Commons, Wikibooks, Wikisource, Wiktionary, Wikivoyage) using the same username and password. If you have third-party cookies enabled in your browser you should find yourself logged in fully automatically on all these projects after signing in to one of them. — Alexis Jazz (talk or ping me) 06:28, 25 May 2021 (UTC)
Yep, the redirect needs to go as it shadows a different file (the 'real' Universo Online logo) at Commons. I've tagged it for CSD G6 as such - of course the tag also 'breaks' the redirect, fixing the problem in the meantime (you might need to purge cache to see the change). ƒirefly ( t · c ) 07:51, 25 May 2021 (UTC)
Anyone else seeing bizarre changes in how Wikipedia displays (since this morning, when all was fine)? Everything that should be small is huge (short descriptions, page data, tab labels, text in edit window etc), while article text is minute, and references even smaller. Vector skin, before you ask. Justlettersandnumbers (talk) 19:22, 20 May 2021 (UTC)
Yes. Something has changed very recently. In particular, categories at the bottom are now larger, and every article has a subheading (or by-line?) "From Wikipedia, the free encyclopedia" in an incongruous font. Certes (talk) 19:30, 20 May 2021 (UTC)
Johan (WMF), you wrote here that a new version of MediaWiki would be on all Wikipedias from 20 May. Can you kindly tell us who gave you that information, and perhaps ask them to respond here? Wikipedia has been rendered virtually unusable without a powerful magnifying glass; if it's as a result of that version change let's hope whoever did it knows where to find the Undo button. Many thanks, Justlettersandnumbers (talk) 20:05, 20 May 2021 (UTC)
Same for me. I initially thought it was browser's zoom level, but it's 100% in my Firefox. I hope it will be fixed soon. Brandmeistertalk20:43, 20 May 2021 (UTC)
Yes, I'm happy yo help, but I don't have any special insight into most minor changes (that can still cause big problems). I see it's been triaged in the linked Phabricator ticket now.
Noticed it right away myself when I logged in this afternoon. I wondered if this were some update that had been in work for some time, but from the looks of things it's at least a bug or it's a new feature rollout that hasn't gone quite right. And yes, I checked my Firefox's zoom level too; it's not uncommon for me to accidentally tweak it, being that I work on a laptop. Zeke, the Mad Horrorist(Speak quickly)(Follow my trail)20:59, 20 May 2021 (UTC)
Yeah, something happened a few hours ago - the font size shrank considerably. Note that case/ticket #T283281 has been closed as a duplicate of T283206, so I updated the number in the Phabricator template. PKT(alk)22:35, 20 May 2021 (UTC)
Hello Whatamidoing - to be honest, your questions are very technical to me. What I can tell you is that, in my case, it's on two different PC's - my work machine and my personal one, and I use Chrome on both. I changed nothing ('cuz I don't understand skins and that sort of thing). Oh - and font sizes on other tabs and sites have not changed. PKT(alk)22:51, 20 May 2021 (UTC)
One more thing to check, @PKT. Please go to Special:Preferences#mw-prefsection-gadgets and look for an item (about 25 lines from the end of the very long page) that says "Vector classic typography (use only sans-serif in Vector skin)". Is that one checked or unchecked?
Thanks for following up @Whatamidoing (WMF):. The font sizes are now larger than they were before the change that took place Thursday, but it's easier to read than the mini-font. However, unchecking "Vector classic typography..." doesn't help the issue with coordinates conflicting with infoboxes, and categories are in a font size that's even larger than before. PKT(alk)10:46, 22 May 2021 (UTC)
@PKT, thanks for checking. It looks like this bug is in the process of being fixed. I don't know whether it will be fixed as an "emergency" or if we'll have to wait until the next normal round (usually Thursdays, unless something breaks). Whatamidoing (WMF) (talk) 17:39, 24 May 2021 (UTC)
I see the same in a private window, even when logged out: large font for categories and the "From Wikipedia, the free encyclopedia" line. For me it's perfectly usable, just odd. I use Firefox 88.0 (86.0 also tested) on the slightly outdated Ubuntu 16.04. I don't have "Vector classic typography" selected: is that the same thing as Edokter's gadget? Safemode brings a different set of oddities: the top-right panel (infobox, or Wiktionary box on a dab) appears badly formatted top left instead of the "From Wikipedia..." line. Certes (talk) 23:01, 20 May 2021 (UTC)
I'm seeing it in Chrome for Mac OS while logged out and in Firefox for Mac OS while logged in, using the "Use Legacy Vector" setting in Preferences/Appearance. Title coordinates are also in the wrong place in Vector, which is a possibly related bug. – Jonesey95 (talk) 23:26, 20 May 2021 (UTC)
I came here to report coordinates being misplaced as well as category font sizes. The coords are being superimposed onto infobox images. Using Chrome on Mac OS also. --DB1729 (talk) 04:29, 21 May 2021 (UTC)
Nice, the choice between ugly fonts or extremely small fonts. Problem is indeed (in human language, not the .css references above) in Preferences, tab "Gadgets", section "Appearance", checkbox "Vector classic typography (use only sans-serif in Vector skin)". If this is checked, the font is since yesterday evening extremely small. If this is unchecked, you get the ugly title font and space-consuming body layout. I have made screenshots of three versions; how it looked like if you had the chackbox marked, until yesterday (good!), and the two poor choices you get now (either way too small, or a lot less on a screen than it used to be). So please, revert this change! Fram (talk) 07:17, 21 May 2021 (UTC)
Old, good version (checked)
New version, checked (small!)
New version, unchecked (large and ugly!)
Thanks, Fram. Took me a while to find that – for others: it's in the Appearance section of the Gadgets tab, not in the Appearance tab. Now everything is ridiculously over-sized. WMF, please restore the status quo ante. As K/Tommy Lee Jones says in MIB 2, "This one is an example of 'go home and do it again'." Thanks, Justlettersandnumbers (talk) 09:11, 21 May 2021 (UTC)
Oh, thanks! I have edited my previous post to add the tab page you need to look at, without it this was indeed confusing. Fram (talk) 09:19, 21 May 2021 (UTC)
This is apparently not the only thing messed up by the recent change. After "User contributions" (on, um, User contribution pages) there's now additional text that reads " for [Username]", the same information that's already present in the next line down, which starts "For [Username]" and gives various links. What's the point of duplicating that information? Especially when in the new over-sized font it makes the page heading absurdly long for long usernames. Yesterday I happened to be looking at Special:Contributions/2A02:C7E:120F:700:74BA:8414:ACF7:73B4, and stupidly reloaded the page. Now the page header is 1720 pixels wide and 66 pixels high. I have a small but fairly high-resolution screen; the text "User contributions for 2A02:C7E:120F:700:74BA:8414:ACF7:73B4" stretches more than two-thirds of the way across it. It looks roughly (not precisely) like this:
User contributions for 2A02:C7E:120F:700:74BA:8414:ACF7:73B4
For the benefit of those of us who cannot easily make sense of Phab entries, can someone tell us if these problems (especially fonts and coordinates placement) are scheduled to be fixed (and when) or is more investigation and/or decision-making required? Or, can we completely solve the problems with routine changes to our Prefs? Do we just need to be patient until next Thursday, perhaps? Nurg (talk) 22:45, 23 May 2021 (UTC)
Looks like the problem has been resolved. The font sizes have returned to normal for me here on the enwiki (although the text padding in previewing from source seems different now). SWinxy (talk) 19:09, 24 May 2021 (UTC)
I am still experiencing small font size on chrome browser, when logged in with my custom css file for vector and 'Use Legacy Vector' checked in preferences (User:Dialectric/vector.css). I could adjust my css to get back to a normal font size, but it was fine until something changed on wikimedia's end. Logged out, the fonts do appear to be fixed. Dialectric (talk) 14:57, 25 May 2021 (UTC)
Large font for categories in articles
Starting today, the text in the box of categories at the end of articles seems to be a larger font. I'm using Vector skin, and that text is now larger than the article prose or the toolbox and other WP interface text. Who did what this Thursday? DMacks (talk) 23:15, 20 May 2021 (UTC)
Having noticed this too, I'm happy to see that it was flagged as a bug rather than being an "improvement" the developers pushed on us without discussion. – wbm1058 (talk) 13:26, 21 May 2021 (UTC)
Looks like it is a Wikimedia thing. If I understand the above and the phab comments correctly the problem has been found but needs to be deployed with the next Wikimedia software update, but there is a temporary fix that the site admins can apply. Nothing you or I can do. — Jts1882 | talk11:52, 21 May 2021 (UTC)
Doubly applied margin
In the past day or so it appears that a margin is being applied twice to all of the articles I read, making them very narrow. The issue doesn't appear in safe mode, and appears to be a result of "removal of the "mw-body-content" class from the "bodyContent" block and its addition to the "mw-content-text" block instead" mentioned by Paul_012 above (manually changing some classes around fixes it). Has anyone else had this issue, or any suggestions other than looking through all of my scripts? LittlePuppers (talk) 13:20, 21 May 2021 (UTC)
If you use the Vector skin, see if you have "Use Legacy Vector" checked under Preferences → Appearance. The beta version of Vector currently causes very wide margins if you have a wide screen. – Rummskartoffel (talk • contribs) 13:50, 21 May 2021 (UTC)
@LittlePuppers, can you check this in a private/incognito window (which is a quick way of logging out in one window, without having the hassle of logging back in afterwards), or in mw:safemode? If the problem disappears in these tests, then we'll know that the problem is in your account. (If it's still there, then the problem is likely something that you can't fix yourself.) Whatamidoing (WMF) (talk) 04:46, 22 May 2021 (UTC)
Whatamidoing (WMF), it does not appear when logged out, or in safe mode (as mentioned above - although safe mode still doesn't look quite right, but that's a different issue, and again, not present when logged out). I suppose I'll begin to look through my scripts then. LittlePuppers (talk) 22:26, 22 May 2021 (UTC)
If by redacted edits you mean oversighted edits, no. That change occurred a week or two ago and was deliberate. Izno (talk) 03:13, 22 May 2021 (UTC)
Category list font-size is bigger
Has anyone else noticed a jump in the font-size of the Category list at the bottom of every article page? It looks about 150% or 200% of what I remember. Looking at one page as an example (Zazou), I'm seeing this html:
<div class="printfooter">Retrieved from "<a dir="ltr" href="http://en.wiki.x.io/w/index.php?title=Zazou&oldid=1024460845">http://en.wiki.x.io/w/index.php?title=Zazou&oldid=1024460845</a>"</div></div>
<div id="catlinks" class="catlinks" data-mw="interface"><div id="mw-normal-catlinks" class="mw-normal-catlinks"><a href="/wiki/Help:Category" title="Help:Category">Categories</a>: <ul> (a bunch of categories are here...) </ul></div><div id="mw-hidden-catlinks" class="mw-hidden-catlinks mw-hidden-cats-user-shown">Hidden categories: <ul><li> (a bunch more here...) </li></ul></div></div>
</div>
</div>
Now the font size looks ok to me, as previously (Firefox 88.0.1, zoom level 100% on a Windows 10 laptop). Not too small as it was. Thanks. Brandmeistertalk21:17, 24 May 2021 (UTC)
I am still experiencing small font size on chrome browser, when logged in with my custom css file for vector and 'Use Legacy Vector' checked in preferences (User:Dialectric/vector.css). I could adjust my css to get back to a normal font size, but it was fine until something changed on wikimedia's end. Logged out, the fonts do appear to be fixed. Dialectric (talk) 14:57, 25 May 2021 (UTC)
Known issue caused by performance work that either can't be fixed or simply was not. AIUI Xtools provides a similar view, so I think it was a time and not feasibility problem. Previously tracked at phab:T282557. Izno (talk) 15:02, 14 May 2021 (UTC)
@Praxidicae @Izno @Nick Moyes As I'm sure you've noticed, GUC is working again and faster than ever! But I wanted to let you know that I did get proper IP range support added to XTools. I could use a hand with testing, if anyone is interested. It's live now on the staging server, i.e. https://xtools-dev.wmflabs.org/globalcontribs/ipr-174.197.128.0/18. See also IP range support in the Edit Counter and everywhere else in XTools! Only the Pages Created tool is lacking support (for now). If you're wondering why the URL has the IP ranges prefixed with ipr-, it was a necessity due to how the routing system works (otherwise it could mistake the /18 as referring to namespace with ID 18). Hopefully that's not too confusing for users, because you can enter normal CIDR notation in the form at https://xtools-dev.wmflabs.org/globalcontribs and it will do the conversion for you. Here on the wiki, we have a separate interface message for IP ranges, so we can link to Global Contribs for IP ranges easily. Templates that link to XTools may require some special handling, though. — MusikAnimaltalk05:03, 21 May 2021 (UTC)
download dialog has disappeared from Chrome mobile. can't rename files, or change save location..would appreciate the advice..many thanks..Gfigs (talk) 20:22, 25 May 2021 (UTC)
Once again, Gfigs, nobody on the planet (except you) has any idea what you're talking about, what you want or what you are seeing. You say you'd appreciate the advice; mine is: give us a clue, maybe two, even. Otherwise, this will just be another thread where you talk to yourself. — JohnFromPinckney (talk / edits)02:37, 26 May 2021 (UTC)
hi, @JohnFromPinckney:, my apologies..what is "Azerbaijan: MTN (until 2015)" ? am guessing it is the equivalent of NSA ? is there perhaps an article on "Azerbaijan: MTN", does not seem to be appearing in MTN ? and if not, why is it appearing in search results, and not in the articles ?..thanks.. Gfigs (talk) 03:12, 26 May 2021 (UTC)
Please write in complete sentences, and then make sure to ask a coherent question using one or more complete sentences. As it is, you are wasting the time of hundreds of editors. – Jonesey95 (talk) 04:57, 26 May 2021 (UTC)
reference when the copyright is not held by the authors
I have just used the reference Navy Board Ship Models [1] - but the copyright information at the front of the book clearly says "Copyright National Maritime Museum, Greenwich 2018". There seems to be no field for this important piece of information. Have I missed it, or does it not exist? Thanks, ThoughtIdRetired (talk) 15:03, 26 May 2021 (UTC)
It's not important for a citation if you have known publisher and author. Sometimes if you don't have one of those the copyright notice can be used to infer one or the other, but that is not the case here. Izno (talk) 15:37, 26 May 2021 (UTC)
As Jts1882 said. More is needed. Try searching for "mw.html.create" in module namespace. There is a page for Lua questions: WT:Lua. Johnuniq (talk) 09:55, 22 May 2021 (UTC)
You need to declare a text variable before (outside) the for loop and append each row to it. Return that variable after (outside) the for loop. As of this version, you are returning the first row within the loop, so it only loops once then exits. If you need further help, WT:Lua can give more relevant advice. Certes (talk) 10:45, 22 May 2021 (UTC)
Got it! I hadn't realised return would break the loop. Now I'm running into time constraints. So is there really no way of returning results as it goes along, so that we can see how far it goes? I'll ask it at WT:Lua too. Thanks. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 11:33, 22 May 2021 (UTC)
Returning from inside a loop without ending execution of the loop requires coroutines in Lua (and other languages), which are disabled in our installation. Izno (talk) 13:14, 22 May 2021 (UTC)
There is an time report when you preview User:Guarapiranga/sandbox/4, at the bottom of the page under "parser profiling data". Under that, you see that the module besides time would also hit the "Expensive parser function count" limit, as it is currently at 472 out of 500. That is caused by your usage of mw.title.new():getContent, mw.title.new().exists and Module:Redirect's mw.title.new().isRedirect.--Snævar (talk) 20:06, 24 May 2021 (UTC)
Thanks, Snævar. In the Edit source tab, after preview changes? I don't see it bc the server spits out Error contacting the Parsoid/RESTBase server (HTTP 504) (I get that a lot). — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 01:17, 27 May 2021 (UTC)
Guarapiranga If you can't access it in the preview changes window for whatever reason then you can just view the html source of the finished page page and near the bottom you'll find a HTML comment which contains the same information. The first line of the comment is "NewPP limit report". 192.76.8.73 (talk) 01:37, 27 May 2021 (UTC)
There have been multiple failed attempts to log in to your account from a new device
Hi, Everyday now I'm receiving the "There have been multiple failed attempts to log in to your account from a new device" notification, Can this notification be disabled as it's now being used by a sock to annoy me. Thanks, –Davey2010Talk11:41, 27 May 2021 (UTC)
@Davey2010: Please see Special:Preferences, click on 'notifications', and look for the header "Notify me about these events". There should be a list item: "Failed login attempts". You can customize your notification preferences from there. SQLQuery me!12:07, 27 May 2021 (UTC)
Thank you so much User:SQL, Being honest I wasn't expecting something like that to exist hence why I didn't actively search. Anyway thanks again for your help it's much appreciated :), Thanks, –Davey2010Talk12:13, 27 May 2021 (UTC)
Error when attempting to restore article
I've had a request to restore List of firsts in India to an editor's userspace. When I try to restore it with full history, I am consistently getting (over a couple of days of trying) the following error:
Database error: To avoid creating high replication lag, this transaction was aborted because the write duration (6.6316473484039) exceeded the 3 second limit. If you are changing many items at once, try doing multiple smaller operations instead. [802e408b-c4c2-4eff-84ce-bd4d017a979d] 2021-05-26 08:16:25: Fatal exception of type "Wikimedia\Rdbms\DBTransactionSizeError"
I can restore a single revision so that the content can be viewed, but the history is lost. Is there currently a server issue (I couldn't find details of any) or can anyone suggest a workaround? The article has a fairly substantial history but nothing out of the ordinary. Thanks. --Michig (talk) 08:22, 26 May 2021 (UTC)
Michig, you can split the restore operation into smaller chunks. For example, you could select around 500 edits and then restore them. Then do another 500 or so until they are all restored. If you click one check box and then shift click a check box around a 1/3rd of the way through the Special:Undelete/List of firsts in India page this should restore around 500 at a time. This shouldn't then give the database error as you are splitting the operation into smaller chunks and thus not hitting the 3 second limit on each undelete operation. Moving shouldn't give the database issue as from memory undeletion takes longer than simply moving. Dreamy Jazztalk to me | my contributions08:42, 26 May 2021 (UTC)
While the advice about splitting large operations into smaller chunks is good advice, another trick to know is when something times out, often the easiest thing to do is just try it again right away. The first aborted attempt may well have pulled the required indexes into cache and a retry will be much faster. But in this case, since you say it's happening consistently, yeah, splitting it into smaller chunks is the right strategy. -- RoySmith(talk)14:51, 26 May 2021 (UTC)
For what it's worth I've found in the last six months or so that for this sort of operation, redoing the undelete doesn't help anywhere near as often as it used to. Graham8708:40, 27 May 2021 (UTC)
Wasn't the default number of items on WhatLinksHere customizable?
I have customized "Number of edits to show in recent changes, page histories, and in logs, by default" in Special:Preferences#mw-prefsection-rc, but Special:WhatLinksHere now always shows 50 items unless the URL has limit=n. I feel like WhatLinksHere respected that setting and showed the custom number of links, or am I misremembering? Nardog (talk) 12:50, 27 May 2021 (UTC)
mw.loader.using(['mediawiki.util'],function(){varname=mw.config.get('wgPageName').replace("Special:WhatLinksHere/","");mw.util.addPortletLink('p-tb',mw.util.getUrl('Special:WhatLinksHere/')+encodeURIComponent(name)+'?limit=100','100 links here','pt-morelinkshere','100 pages containing links to this page',null,'#t-whatlinkshere');});
I am a Crosstor and I ask if there is an archive that has been in a recorded state for 3 years now, for example? I want to know how my article once disappeared without a trace from Esperanto Wikipedia. All my searches failed. If I had a fixed condition for 3 years, I would definitely find the article. This is not listed on the standard cancellation discussion board, meaning someone has deleted it arbitrarily and secretly. Sincerely, Crosstor. — Preceding unsigned comment added by Crosstor (talk • contribs) 13:36, 27 May 2021 (UTC)
Also the direction say no leading zeroes. (example: 9 - correct, 09 - incorrect, however you used leading zeros. — xaosfluxTalk14:51, 27 May 2021 (UTC)
@Dr Salvus: As PrimeHunter suggested, you should remove the other scripts you have installed above the wikibreak enforcer (that is, lines 1-10). It is not unlikely that an error in one of the other scripts is preventing the execution of the wikibreak enforcer. EDIT: actually, now that I look at it, I don't know where line 11 comes from, but it looks like it will almost certainly cause an error. You should remove that line. Writ Keeper⚇♔15:42, 27 May 2021 (UTC)
Thanks @Rummskartoffel: Thanks. I can't change it there as it's protected. There are other problems with that template which I discovered recently. I don't know enough about templates or the mysteries of en-dashes to try to sort it out. DuncanHill (talk) 15:46, 22 May 2021 (UTC)
I made a naïve first attempt at detecting when to use what here, but that currently breaks with at least {{Gambia topics}}. (I think it may be that template's fault, but something would have to be done about it either way before, if at all, these changes could be used.) – Rummskartoffel (talk • contribs) 21:13, 22 May 2021 (UTC)
Using a redirect would work, but that'd kind of defeat the purpose here. I've noticed, however, that {{Gambia topics}} calls {{Country topics}} with |country=The Gambia instead of |country=Gambia and |prefix=the, so maybe the way to fix this would be to change that instead. – Rummskartoffel (talk • contribs) 22:12, 22 May 2021 (UTC)
Is it me, or has the text whitespace around thumbnails gotten bigger suddenly, especially below them:
Example text
Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
Curabitur pretium tincidunt lacus. Nulla gravida orci a odio. Nullam varius, turpis et commodo pharetra, est eros bibendum elit, nec luctus magna felis sollicitudin mauris. Integer in mauris eu nibh euismod gravida. Duis ac tellus et risus vulputate vehicula. Donec lobortis risus a elit. Etiam tempor. Ut ullamcorper, ligula eu tempor congue, eros est euismod turpis, id tincidunt sapien risus a quam. Maecenas fermentum consequat mi. Donec fermentum. Pellentesque malesuada nulla a mi. Duis sapien sem, aliquet nec, commodo eget, consequat quis, neque. Aliquam faucibus, elit ut dictum aliquet, felis nisl adipiscing sapien, sed malesuada diam lacus eget erat. Cras mollis scelerisque nunc. Nullam arcu. Aliquam consequat. Curabitur augue lorem, dapibus quis, laoreet et, pretium ac, nisi. Aenean magna nisl, mollis quis, molestie eu, feugiat in, orci. In hac habitasse platea dictumst.
I'm guessing this is one of those "perennial" questions. Could someone point me to some previous discussions that ask about separating article/user pages from the talk pages for the watchlist. Example: I want to watchlist John Wayne, but don't want to see the talk page show up in my watchlist (or the reverse). I had asked on VP talk, but it seems to be rather sparsely tended to at the moment. If this belongs at a different section, (Idea lab, proposal, etc.), feel free to move. thanks. — Ched (talk) 07:39, 27 May 2021 (UTC)
@Ched: there is no option that we could do here on the English Wikipedia for this, it would require upstream software work. You could refresh and reopen phab:T12175 (from 2007) about this to at the least get a more documented "Decline" response! — xaosfluxTalk10:46, 27 May 2021 (UTC)
Kusma: That's bloody brilliant! I had no idea that was an option. Maybe I should spend some time looking at the Special pages. Thanks a bundle,— JohnFromPinckney (talk / edits)12:35, 27 May 2021 (UTC)
First, thank you very much for the prompt responses. I had completely forgotten about that related changes option. Kusma, thank you for that memory jog - I think it's been 10 years since I heard about it (and never did act on it). It is most definitely something I'm going to set up in the next few days. I agree with John - "bloody brilliant". One thing I'll check is to see if the {{noping}} works in that setup. Xaosflux, thank you for finding that (I honestly don't know how some folks find things so old and obscure sometimes). I bookmarked it, and may dig up that "wish list" thing that meta/media has, and add/request it to something for developers to work on. TY again. — Ched (talk) 15:17, 27 May 2021 (UTC)
The call to {{tl:X24}} results in "Coordinates: Missing latitude", even though the two templates should do the same. Any idea why this is happening? Thnx, Ponor (talk) 06:47, 28 May 2021 (UTC)
I believe it's because {{Template:Coord}} is specifically identified as a wrapper template in Module:Coordinates (line 54). See Module:Arguments § Wrappers for an explanation. Basically, to get it to work like you want, you'd have to add {{Template:X24}} to the wrappers list in Module:Coordinates — sbb (talk) 07:31, 28 May 2021 (UTC)
... that is to say, as currently coded, Module:Coordinates can only be invoked like that from{{Template:Coord}} — sbb (talk) 07:35, 28 May 2021 (UTC)
Is there a way to automatically repopulate a restored category? (Or populate it based on another wiki)
@Piotrus: no, the only way to "populate a category" is to make an edit to each page that will be in it. This could be semi-automated with tools if there is a specific list of pages known. — xaosfluxTalk11:08, 28 May 2021 (UTC)
@Piotrus, you could go and check who did the removal and work from their contributions list. Might be possible to use AWB to turn that into a list of pages that AWB then can add the categories to (haven't used AWB in 10 years, though). —Kusma (talk) 11:29, 28 May 2021 (UTC)
I've found the bot that did the removal work and have managed to create a list of all the articles it was removed from (See wall of text here). If you click edit, the list will be easier to read. Check it over and if it looks OK, make a request at WP:AWBTASKS for someone to use the list to restore the category to those articles. - X201 (talk) 11:43, 28 May 2021 (UTC)
Hello everyone,
My apologies for writing that on en.wiki, but I don't know which Wikimedia-website to consult for this kind of problem. The wa.wiki main page is playing up right now for unknown reasons. I precise that nobody has modified the "main page" (which is, of course, protected) nor a template/module related to. Wa.wiki uses the same module of main page as wa.wiktionary. However, the main page of wa.wiktionary seems still working correctly.
Thank you in advance for your help,
--Èl-Gueuye-Noere (talk) 18:31, 29 May 2021 (UTC)
Yes, it was this one. I finally found the bug, it was a thing like this : < / div>, which added itself. That's why, the main page bugged without anyone nor any bot modifies it. I don't know if this kind of bugs often happens. I was afraid that something worse had happened. --Èl-Gueuye-Noere (talk) 19:23, 29 May 2021 (UTC)
Hmm, looks like this is by design via the title blacklist. I've taken out the prompt to direct IP editors to the ACW for redlinked categories. You can follow up at Wikipedia talk:Article wizard if you would like to discuss if this should be redesigned. — xaosfluxTalk21:08, 29 May 2021 (UTC)
Consider the pages Wikipedia:VPT and Wikipedia:Village Pump (technical). Here, Wikipedia:VPT actually redirects to Wikipedia:Village Pump (technical). My question is, is there any method by which I can identify all the pages which contains the text Wikipedia:VPT that is actually a redirect? I have checked What Links Here, but its not a solution as it contains links with both the source and redirected title. Adithyak1997 (talk) 10:28, 30 May 2021 (UTC)
@MusikAnimal: thanks very much for being so helpful! 👍 I appreciate you not sending me offsite over something so small. Vive NOT:BURO! All the best. ——Serial10:58, 30 May 2021 (UTC)
Save changes only where necessary, instead of (as some people do) making one big edit followed by four or five small ones the only purpose of which is to correct the spelling and grammar. --Redrose64 🌹 (talk) 20:36, 28 May 2021 (UTC)
I would like to see WMF publish statistics on power consumption. Big data center operators like Google, Facebook and AWS put a lot of effort into reducing power consumption. From their point of view, it's about saving money, but reducing the electric bill and reducing the carbon footprint are closely aligned. WMF publishes some stats, but I don't see anything about power consumption. Computing and publishing J/view (Joules consumed per page view) would be a good place to start. What you'd like to see is a steady decline over time. Since WMF doesn't run their own data centers, this can get a bit messy to figure out, but you gotta start somewhere. -- RoySmith(talk)21:13, 28 May 2021 (UTC)
You might be interested in m:Sustainability which has lots of data and reports on this. For example in 2019 data centres were reported to be using an average of 10.78 kWh and emitting 2.3 kg of CO2 equivalent per million pageviews. It's also worth noting that a significant part of Wikimedia's carbon footprint comes from staff and community travel to conferences such as Wikimania. Obviously this past year will have seen a significant cut in that! I'm interested to see how well Wikimania 2021 works as a fully remote event, and how we can increase remote participation at events in future. the wub"?!"22:02, 28 May 2021 (UTC)
At http://en.wiki.x.io/w/index.php?title=User:Xaosflux/sandbox101&diff=prev&oldid=1018149200 (and any other suppressed revision), when you press on the top revision text (that says Revision as of 14:33, 16 April 2021 in this one), it says that this page revision has been removed from the public archives. Details can be found in the deletion log for this page. It should say that "this page revision has been suppressed", as it is oversight, not revision deletion. aeschylus (talk) 12:46, 30 May 2021 (UTC)
@Ivanvector, it wasn't as I saw one suppressed one and one deleted one, and the deletion log only shows one revision being deleted, so it wasn't revdeleted first. aeschylus (talk) 16:45, 30 May 2021 (UTC)
So, depending on your access levels you can come upon: MediaWiki:rev-deleted-text-unhide (for revdel only if you are an admin), MediaWiki:rev-suppressed-text-unhide (for os if you are an os), but if you have no access you do end up at MediaWiki:rev-deleted-text-permission. Now you end up at that last on regardless of why you can't see it and in the overwhelming majority of cases it will be because of normal deletion. The default prompt for that message is:
This page revision has been <strong>deleted</strong>.
Details can be found in the [{{fullurl:{{#Special:Log}}/delete|page=$1}} deletion log].
So we haven't materially changed the message and it doesn't have support for a conditional variable based on if suppression is also applied, and likely doesn't really need it for non-administrators - if you want support for that you would need to file a software request on that page logic. — xaosfluxTalk13:49, 30 May 2021 (UTC)
@Aeschylus:WP:BUG has the link to Phabricator. It is likely to go in to the pile of not-declined-but-no-one-will-work-it on like other very low priority things that have an over 10 year backlog - as you don't seem to have identified something that is going to impact readers or prevent editors from working. — xaosfluxTalk17:06, 30 May 2021 (UTC)
Template:Infobox Site of Special Scientific Interest
There seems to be an image caption positioning issue in {{Infobox Site of Special Scientific Interest}} which can be seen in Bewick and Beanley Moors, and which is that when there is also a locator map, the caption for the image is beneath the map, rather than beneath the image. The dissociation of image and caption seems wrong and unhelpful. Grateful if this issue could be given some attention. --Tagishsimon (talk) 20:34, 30 May 2021 (UTC)
For me, tables with the class "mw-collapsible" or using {{collapse}} (or similar), instead of having a [hide] link, it appears like this: [hide]. ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 21:12, 30 May 2021 (UTC)
@Qwerfjkl Works for me. Can you maybe open the developer console to see if there are any problems loading any of the scripts? Anything with "ERROR" or whatnot. Aasim (talk) 23:11, 30 May 2021 (UTC)
@Qwerfjkl Do the tables still collapse? Maybe if you paste a link to a screencast showing the problem I could maybe diagnose it. Aasim (talk) 07:27, 31 May 2021 (UTC)
It is more than likely you have a conflicting set of scripts as well... would like to see a screencast or whatever... it will help me identify the problem. Aasim (talk) 07:28, 31 May 2021 (UTC)
Is there a way to mark individual entries as read in the watchlist?
I'm an avid user of Writ Keeper's excellent In-place diffs, especially in the watchlist, so I don't always need to open the changed page. It'd be great if I could cross that off the watchlist rather than keep count of everything I already looked at before clicking "mark all as read". I see this was proposed 10 years ago but couldn't find the solution. I would've thought that bullet in the watchlist could work as read/unread toggle, as it does under notifications. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 00:32, 31 May 2021 (UTC)
phab:T92947 is about the inverse of this, but has some suggestions that this could possible be done with API calls, if so a userscript could be written for it. — xaosfluxTalk01:18, 31 May 2021 (UTC)
Hmmm, well, it should be pretty simple to make a quick script that will do this on the standard watchlist; I'll whip something up uf you're still interested. Writ Keeper⚇♔08:32, 31 May 2021 (UTC)
Awesome!, Writ_KeeperIt works. How does one unread the revision though (if that's what you meant)? When I click the link/button, it disappears (but not the entry). The script works well enough, but if you feel like polishing it, I'd suggest striking through the entry marked as read (and unstriking if you manage to mark it as unread), and—if at all possible—using the far left bullet as button instead of adding another, like UncleDouggie did. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 09:47, 31 May 2021 (UTC)
I'll do a proper debug when I'm back at the desktop, but I'm happy to report that it's working on my mobile (and also that my suggestion of using the far left bullet wouldn't work on mobile as it's not there). Thanks. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 10:59, 31 May 2021 (UTC)
Re-review this revision—from 2017
I'm used to seeing notices about "configured pending changes settings" in history, but never boxed up and presented at the bottom of the article page below the categories like this. So why does Trypophobia have a box echoing a revision made by Callanecc in 2017, and formatted just like a review pending changes input box, with a button for me to Unaccept their years-ago edit? Was this part of the release that gave us the mystery-font size issue?
Snippet of html from the footer of Trypophobia
<divid='mw-data-after-content'><formmethod="post"action="/w/index.php?title=Special:RevisionReview&action=submit"id="mw-fr-reviewform"><fieldsetclass="flaggedrevs_reviewform noprint"><legendid="mw-fr-reviewformlegend"><strong>Re-review this revision</strong></legend><p></p><pclass="fr-rating-controls"id="fr-rating-controls"><spanid="mw-fr-ratingselects"class="fr-rating-options"></span><spanid="mw-fr-confirmreview"><labelfor="mw-fr-commentbox"class="fr-comment-box">Comment:</label><inputname="wpReason"size="40"value=""id="mw-fr-commentbox"maxlength="500"class="fr-comment-box"/><inputname="wpApprove"id="mw-fr-submit-accept"accesskey="s"title="Mark this revision as accepted [s]"disabled=""type="submit"value="Accept revision"/><inputname="wpUnapprove"id="mw-fr-submit-unaccept"title="Revoke acceptance of this revision by marking it as unaccepted"style=""type="submit"value="Unaccept revision"/><script>varjsReviewNeedsChange=0;</script><spanid="mw-fr-logtoggle"class="fr-logtoggle-excerpt"style="display:none;">(<aclass="fr-toggle-text"title="Toggle display of the latest entry in the pending changes protection log">hide pending changes protection log</a>)</span></span><divid="mw-fr-logexcerpt"><ulclass='mw-logevent-loglines'><lidata-mw-logid="86418828"data-mw-logaction="stable/config"class="mw-logline-stable"><ahref="/w/index.php?title=Special:Log&logid=86418828"title="Special:Log">05:02, October 21, 2017</a><ahref="/wiki/User:Callanecc"class="mw-userlink"title="User:Callanecc"><bdi>Callanecc</bdi></a> configured pending changes settings for <ahref="/wiki/Trypophobia"title="Trypophobia">Trypophobia</a> [Auto-accept: require "autoconfirmed" permission] <spanclass="comment">(Persistent addition of <ahref="/wiki/Wikipedia:INTREF"class="mw-redirect"title="Wikipedia:INTREF">unsourced or poorly sourced content</a>)</span><spanclass="mw-logevent-actionlink">(<ahref="/w/index.php?title=Trypophobia&action=history&offset=20171021050237"title="Trypophobia">hist</a>)</span></li></ul><ahref="/w/index.php?title=Special:Log&page=Trypophobia&type=stable"title="Special:Log">View full log</a></div></p><inputtype="hidden"value="Special:RevisionReview"name="title"/><inputtype="hidden"value="Trypophobia"name="target"/><inputid="mw-fr-input-refid"type="hidden"value="0"name="refid"/><inputid="mw-fr-input-oldid"type="hidden"value="1026062748"name="oldid"/><inputtype="hidden"value="3928ec26fb9fdebbabbead26754c8eaf60b470c3+\"name="wpEditToken"/><inputid="mw-fr-input-changetime"type="hidden"value="20210531050433"name="changetime"/><inputid="mw-fr-user-reviewing"type="hidden"value="0"name="userreviewing"/><inputtype="hidden"name="templateParams"/><inputtype="hidden"name="imageParams"/><inputtype="hidden"name="fileVersion"/><inputtype="hidden"value="d9d7ef2024098573ee57c1de169900a8"name="validatedParams"/></fieldset></form><divclass="read-more-container"></div></div>
I think it's not about unaccepting Callanecc's revision, but your latest auto-accepted revision? (It seems to happen for all Special:StablePages) But I agree that this interface should not be on the article page, especially as it contains the exact wording of the log entry, which isn't usually written for the general public. There seems to be a "hide" link, but I don't know whether that's admin only and when we should use it. Are there instructions for this anywhere? Wikipedia:Pending changes doesn't help all that much. (I hate pending changes). —Kusma (talk) 09:45, 31 May 2021 (UTC)
Thinking about it, I can't see a good reason to unaccept a page without looking at the page history. Why would you unaccept a revision without checking whether the previous revision is any better? This interface should not be visible on the page itself. —Kusma (talk) 11:19, 31 May 2021 (UTC)
Why is Script Installer still using importScript?
Just curious... If importScript is deprecated and mw.loader should be used instead,[1] why is Script Installer still using importScript? Is it just a string replace in the script, or does it require some transition? Cheers. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 00:19, 27 May 2021 (UTC)
What are the advantages of using mw.loader over importScript? Would it cause any issue if all user scripts were mass-updated to use mw.loader over importScript? Headbomb {t · c · p · b}01:10, 27 May 2021 (UTC)
importScript is just easier to use. Also see T36958, which is what we're waiting for before actually hard deprecating importScript and telling people to move away. Legoktm (talk) 02:32, 27 May 2021 (UTC)
I don't see that task as blocking it at all? It's already 'hard deprecated' given that it emits console warnings. Izno (talk) 18:14, 27 May 2021 (UTC)
mw.loader is better because it's async, but as Legoktm alluded to, it has a dreadful "user interface" (i.e. you have to give it a full URL, which is difficult to do by hand). I guess that's not as important of a concern these days. But I'd still like it if we created an async function with importScript's "user interface" - i.e. you'd only need to give it a page name - and migrated everyone to that. Now, to actually answer the question, I figure switching Script Installer over might as well be done together with any migration we do in the future (which will also involve changing the docs, other script tooling, etc). Enterprisey (talk!) 05:20, 27 May 2021 (UTC)
@Enterprisey I'm having trouble understanding why people are saying importScript is not async. Multiple importScript calls do take place simultaneously. So what's the difference? – SD0001 (talk) 06:58, 28 May 2021 (UTC)
If anyone's interested, I've got a way of loading user scripts that's much superior to the existing methods, see User:SD0001/Making user scripts load faster. It speeds up script loads 100x (that is, if you have a below-average connection like mine) by caching them. Enjoy! – SD0001 (talk) 06:55, 28 May 2021 (UTC)
Ok, First thing that it did, having installed it through my mobile, was to make more scripts run on it (before only the first one or few were, I think). — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 21:40, 28 May 2021 (UTC)
I disabled RedWarn, and kept your script cache. It does load scripts a lot faster than without. Can't it be made into and loaded as a common user script, SD0001? — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚 (talk) 23:46, 30 May 2021 (UTC)
@Guarapiranga The way it works is by over-writing the definitions of importScript, mw.loader, etc. So if it's loaded as a user script, the network requests for fetching other scripts would have already started by the time this one loads, so it'll have no effect ;) But you use the minimised version instead to save lines in common.js (I updated the page to that effect). – SD0001 (talk) 16:22, 31 May 2021 (UTC)👍
@SD0001: Have you tested that? According to the MDN, eval() is also slower than the alternatives, since it has to invoke the JavaScript interpreter, while many other constructs are optimized by modern JS engines.. I don't think new Blob() has that problem, though of course both will raise the blood pressure of anyone looking at your code. Suffusion of Yellow (talk) 20:03, 28 May 2021 (UTC)
I think the bottleneck here would be the fetch rather than the execution, esp. for those of us on slow connections. A normal fetch with index.php takes 300-500 ms for me. But with caching, most of the time the file gets retrieved from Chrome's disk cache which is like 3-5 ms. I suspect modern browsers execute code very fast, so the minor slowness of eval shouldn't be an issue. – SD0001 (talk) 04:52, 29 May 2021 (UTC)
I've brought this up at Template talk:Install user script as well and have been shot down. I personally don't see any reason why either that template or the Script Installer need to be installing non-https scripts or installing scripts via a relative path, so would think both could be migrated to use ResourceLoader mw.loader.load. --Ahecht (TALK PAGE) 16:36, 30 May 2021 (UTC)
The excuse used at the fab ticket linked above (T27845) for why mw.loader.load was not a good substitute for importscript was that you had to specify whether or not to use https when using mw.loader.load. --Ahecht (TALK PAGE) 01:37, 31 May 2021 (UTC)
This basically, though it should be encodeURIComponent(location) instead of just location (otherwise it'll fail on loading pages with names having a space in them). – SD0001 (talk) 16:27, 31 May 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Problems
There was an issue on the Vector skin with the text size of categories and notices under the page title. It was fixed last Monday. [4]
I'm just a poor back-end server guy, who occasionally dabbles in javascript, usually with sub-standard results. I've got two scripts (source here) I've written which appear to have a race condition which sometimes causes them to run twice. I load them from User:RoySmith/common.js. The first:
does some DOM-manipulation to {{checkuser}} templates to add an indication of how the account is tagged. That's the "M" in the yellow box in the screenshot. The other ones:
add the "SPI Tools" and "SPI Tools (dev)" items to the More menu. Maybe once in 50 page loads, one or both of them runs twice. The screen shot shows an example where they *both* ran twice. Anybody know what's going on? I assume this is not using the mw.hook() function correctly? Perhaps wikipage.content is the wrong event to be using? -- RoySmith(talk)16:24, 31 May 2021 (UTC)
If the script only needs to run once per page load, you're better off using $.ready.then(function() {...}); rather than hooking on wikipage.content – which is meant for scripts that need to run following changes to the DOM caused by other javascripts running on the page. – SD0001 (talk) 16:37, 31 May 2021 (UTC)
SD0001, Thanks. Now I just need to figure out how $.ready.then() differs from the
It doesn't differ. $.ready is documented here. No idea why jQuery has so many syntaxes for the same functionality. – SD0001 (talk) 20:08, 31 May 2021 (UTC)
SD0001, Well, there's only so many pieces of punctuation. It would be a shame if they didn't make every possible combination do something, wouldn't it? Thanks again for your help. -- RoySmith(talk)23:02, 31 May 2021 (UTC)
I am seeing hundreds of error messages on COVID-19 pandemic in India. The error is in bold and red "Lua error: not enough memory." When trying to edit the page, more errors of the same kind appear. (Also seeing the same at the commons category for the same article "Category:COVID-19 pandemic in India") DTM (talk) 12:45, 26 May 2021 (UTC)
It's not about the raw size of the article, but rather the amount of time the various modules transcluded onto the page take to execute. You might get some joy asking here for assistance streamlining things. firefly ( t · c ) 13:23, 26 May 2021 (UTC)
This doesn't seem to be a Lua problem or due to article size. The error was introduced by this edit on 24 May 2021. The critical part was removing {{Update|date=April 2021}}. I've put it back and all those Lua errors they have gone. They can also be removed by removing the date preference template so there is some weird interaction of the templates above the infobox. I've no idea why. I've left the unnecessary update template as I think that is better than all the red lua errors, but I have no clue about how to fix the problem. — Jts1882 | talk16:55, 26 May 2021 (UTC)
RoySmith I suspect that what is happening here is that some template on the page is pulling information from the associated wikidata entry, which is causing the page to go over the memory limit as the wikidata entry as it stands is a 3MB page with several thousand entries. This would explain why the issue only shows up on the main article - it's the only one linked to the wikidata entry. I can't see what template could be causing it though - the only template I can see on the page that is reading from wikidata is Template:Official website. 192.76.8.73 (talk) 01:27, 27 May 2021 (UTC)
That's definitely it: the /temp page (with no Wikidata item) has Lua memory usage 10,553,034 bytes, the current main page has memory usage 51,449,827 bytes (only a tiny amount less than the limit), and if I preview removing the call to {{coord}} (which relies on Wikidata to populate some tracking categories) from the infobox, it reduces the memory usage to 28,859,423 bytes, and previewing removing the call to {{official website}} in addition reduces the memory usage to 11,431,012 bytes. Memory usage accounting isn't consistent from one parse to another, so that can be considered equivalent to the /temp page. * Pppery *it has begun...01:47, 27 May 2021 (UTC)
Elli Fetching properties from Wikidata is a mess. If you use a function like mw.wikibase.getBestStatements() which is designed to just retrieve a selection of values from a single property of an item the way it works is by first loading the entire item into lua memory, then it filters the properties to keep the single one you asked it to retrieve in the first place. There was a Phabricator task a couple of years back asking for this to be optimised to stop it from loading entire items when you were just accessing data for a single property, but it was closed with no action, on what appears to be the basis that "It's always worked that way". 192.76.8.73 (talk) 02:30, 27 May 2021 (UTC)
Last time I looked into it, I thought that it would be helpful if the template call would specify how many statements it is expecting. If there is only one needed, then the module can skip an loop looking for every one under an specified property.--Snævar (talk) 20:15, 27 May 2021 (UTC)
Snævar, that wouldn't make much if any difference to the memory usage - the fundamental problem is that there is no way to only access specific properties of an item on wikidata - the only way you can access information is to load the entire page into local lua memory, then sort it to pull out what you asked for, which is why cross checking an official website parameter uses 17MB of memory - it has to load the entire 3MB wikidata page into memory then run a sorting algorithm on it. As I see it there are 3 ways to deal with this issue
The developers rework the wikibase/Scribunto intergration. Facilities to search for and retrieve individual properties are built into wikibase, so that accessing a property in Lua returns only the value of that property to Lua memory.
We remove wikidata integration from any pages where the wikidata page gets excessivley large.
Wikidata splits their large entries into several smaller pages, e.g. "Covid 19 deaths in India" "Covid 19 vaccine rollout in India" "Covid 19 cases in India"
This first of these would require significant development time, the second would remove functionality from articles and the third would kind of defeat the purpose of a centralised database. I think for this specific case the best immediate solution is 2, as the wikidata integration isn't really doing anything important (it's only cross-checking a website and populating some categories) but as a long term solution we should probably be looking towards option 1. 192.76.8.73 (talk) 15:44, 29 May 2021 (UTC)
No, it's not quite that, as there are various other uses of Wikidata that don't consume appreciable memory, such as the {{short description}} (via code in Module:SDcat using mw.wikibase.getDescription) and {{commons category}} (via code in Module:WikidataIB using mw.wikibase.getBestStatements). It appears the tracking in Module:Coordinates could be easily re-written to use that function and presumably not use a huge amount of memory, but the same thing does not apply to Module:Official website, which uses some fancy logic to prefer English-language websites over websites in other languages. * Pppery *it has begun...16:18, 29 May 2021 (UTC)
Pppery I don't think either of those templates are actually accessing wikidata properties in normal usage. Short descriptions aren't stored as properties, they're stored as labels and are kept in a separate "terms table" which can be loaded independently of the rest of wikidata entry. Module:WikidataIB doesn't use getBestStatements by default - it first looks for commons categories in the interlanguage links section of the wikidata which again are stored in a different format and can be loaded independently of the main wikidata entry. Phab:T198116 includes a respones from hoo (who wrote the getBestStatements function) which seems to confirm that accessing a wikidata property loads the entire entry into memory. 192.76.8.73 (talk) 17:37, 29 May 2021 (UTC)
Re 192.76.8.73: I dodged that point previously just not to sound rude. The reality is that you are confusing mw.wikibase.getEntity():getBestStatements() with mw.wikibase.getBestStatements(). The former will load the whole item into memory, but the latter will not. Actually, I find that the dev in the bug actually tried to explain this without succeeding. The dev, Hoo, even says in the bug that it "get[s] just the statements we care about". Try several methods on the wiki and see it for yourself, verifying it with "parser profiling data" in the edit window, instead of only reading about it. If there really was no performance improvement possible then I never would have been able to improve the performance of is:Module:WD-gildi, see how the module disappears from this dashboard. Sure, it is an timing improvement, but that still shows my point.--Snævar (talk) 06:50, 2 June 2021 (UTC)
In this case I don't think Wikidata is the cause. It's something to do with an interaction with the four templates above the infobox: {{pp-semi-indef}}, {{Update}}, {{EngvarB}}, and {{Use dmy dates}}. Remove all of them and the page loads without the Lua errors. It was the removal of {{Update}} that revealed all the Lua errors. I put it back to remove the errors but it is unsatisfactory to have template saying the page needs updating when it has been updates. The other templates are required so removing them is not a solution. I've tried various combinations and cannot see which template is the problem. There seems to be a strange interaction, but its beyond me. — Jts1882 | talk14:30, 31 May 2021 (UTC)
It definitely has to do with Wikidata. I suspect the reason why the templates above the infobox are behaving the way they are is code in Module:Citation/CS1/Configuration that parses the page to determine what date format to use, but can't explain exactly why that occurs. Here's a slightly better way to get the page below the memory limit: change |display=inline,title to |display=inline in the {{coordinates}} template (and remove the {{update}}). This removes the coordinates at the top of the page, but it's not clear they should be there in the first place since the pandemic isn't definable to one location. * Pppery *it has begun...15:16, 31 May 2021 (UTC)
I found out that the coord transclusion in the infobox under "first case" uses 22 millon bytes in memory. Submitted an fix in Module:Coordinates which reduces wikidata memory usage and frees the majority of that memory. Tested that by previewing the COVID India page with the Coordinates sandbox, which includes the fix.--Snævar (talk) 07:24, 2 June 2021 (UTC)
Great work! Of course I didn't believe that was possible but checking shows that the current article uses 51 MB of Lua memory, and removing the single {{coord}} ({{coord|10.5276|76.2144|format=dms|type:village_region:IN|display=inline,title}}) reduced that to 29 MB. Johnuniq (talk) 07:57, 2 June 2021 (UTC)
Just removing "title" from the |display=inline,title in {{coord}} as Pppery suggested produces the same reduction from 51MB to 29MB. And as the location applies to the first covid case and not the whole article on covid in India, I don't think it should be there.
That difference between mw.wikibase.getEntity():getBestStatements() and mw.wikibase.getBestStatements() and use of mw.wikibase.entityExists(qid) is worth remembering. — Jts1882 | talk08:32, 2 June 2021 (UTC)
It is. Scripts description pages—when they exist!—are all over the place; some use {{Infobox user script}}, many don't. Some use their own templates, others just simply write some body text. Having said that, I've just now endeavoured to include descriptions in user script infoboxes. Either way, instead of getting it precisely right, the description there is meant to give a first brush of what the script is about before one clicks through to the description page (through the → link), to investigate what it's all about. An alternative is the manual description in WP:USL. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎02:47, 1 June 2021 (UTC)
Ah, I see things have been updated. There's a strange bolding going on, and things not from outside the description field, but it's better than what was there before. Headbomb {t · c · p · b}03:19, 1 June 2021 (UTC)
The name bolding in the description column is coded in the module, mimicking title bolding in articles' text. I was also adding the script name to the beginning based on BrandonXLF's doc template. Now switched to capitalising the first letter instead. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎06:35, 1 June 2021 (UTC)
You might want to do something about the description for User:Xenocidic/statusChanger2.js, which seems to be taking someone's comment from the talk page (after following a redirect). I appreciate the problem though as the location of the description for the scripts seems quite random. — Jts1882 | talk07:03, 1 June 2021 (UTC)
It looks like the description page was created by another user and then redirected somewhere strange and then redirected to the script talk page. I don't recommend using the script (since I have no idea if it still works), but feel free to change the redirect to a more usable description. It's meant to update a page with your online status, quite old school. Really neat script list though, I've had that pain point of searching for obscure scripts. Thank you! –xenotalk08:44, 2 June 2021 (UTC)
What's going on with bulletpointing using the asterisk? Unless it's just me, something has clearly changed, as if to create a new "level zero" of black circular bullet points instead of the grey square ones.
I’m not quite sure what you mean, but if you want a sublist within a bulleted list (<ul>) you should be using ** x (which means <li><ul><li>x</li></ul></li> within the original list, not :* x which I believe will end the original <ul> and create a completely separate definition list containing a bulleted list. User:GKFXtalk22:38, 31 May 2021 (UTC)
This is not about sublists, but the first-level list. Something in Template:ITN candidate, or its subtemplates, is causing a series of <li>...</li> elements to be emitted bare - that is, without the enclosing <ul>...</ul> tags. --Redrose64 🌹 (talk) 22:58, 31 May 2021 (UTC)
Yes, that makes sense, and it's clearly only been happening since that edit. GKFX, can you fix the problem or revert the template changes, please? Black Kite (talk)23:13, 31 May 2021 (UTC)
I was confused because I was viewing it on mobile, and I still can't see any square bullets (might be a difference in skins). Anyway it is now fixed (Special:Diff/1025987272) - it turns out that you shouldn't write * bullet point</div> with no newline before the tag. I still don't quite understand what happened, but it works now. User:GKFXtalk18:59, 1 June 2021 (UTC)
Yes, 'don't write that specific phrase' is a case of Parser.php being Parser.php. Parsoid handles that case correctly, but Parsoid is not yet live for read views. Izno (talk) 16:15, 2 June 2021 (UTC)
Heads up for CSRF token change
I imagine most people this affects have seen it already, but just in case, I'm giving this more visibility. If you run code that uses CSRF tokens, you'll need to know about this change. -- RoySmith(talk)16:17, 2 June 2021 (UTC)
Is there a way to display the total numbers of links in Special:LinkSearch?
Hi all
I've been playing around with Special:LinkSearch, I think it could be an extremely useful for Wikipedians working with external organisations to encourage them to share knowledge and content if we could provide them with a total number of links. Is there something I can add to the URL to make it display a total? Here's an example search
No, it wasn't. You could do it with Quarry or DB access from Toolforge (in the case here, probably something like SELECT COUNT(*) FROM externallinks WHERE el_index_60 LIKE 'https://org.unesco.%';), or with a lot of API queries and some manual processing of the results. A count built into the page would probably have to be capped for performance. Anomie⚔11:52, 2 June 2021 (UTC)
Thanks Anomie and Xaosflux using quarry or toolforge is way beyond my technical skill level (and most others who would want to get this info), unless there were very simply step by step instructions and I just had to change the URL in a preformatted query. Would that be possible? Just to check I understand, it would be possible to have a total number of links in the linksearch but this has been turned off because it would cost too much processing time? Or has it just not been added because no one asked for it? John Cummings (talk) 12:03, 2 June 2021 (UTC)
It hasn't been implemented because it would cost too much processing time for some cases. Specifically, database time scanning the rows to get a count. Anomie⚔12:20, 2 June 2021 (UTC)
I don't think there is any way of getting the total number of external links other than by getting all items from linksearch and counting them. Re the idea of providing an incentive for organizations, that was tried by the Wikipedia Library (WP:TWL) a few years ago and I analyzed some external link dumps for them. The idea is a bit controversial because the plan is essentially to help an organization promote itself on Wikipedia by spamming their links. My analysis scripts were very klunky and someone wrote a much better system with a funky name that I can't find at the moment. It was a website where a configuration file told some software what domains (like example.com) to monitor, and it would track all external links (like www.example.com/click/bait.html) and show totals over time. I have a feeling that the website shutdown but don't know. I have found something in my talk archives mentioning the "most referenced domains on the English Wikipedia (2015-05-15)" at phab:P587. Johnuniq (talk) 11:34, 2 June 2021 (UTC)
Thanks for the explanation Johnuniq, I didn't mean to use this number as a way of adding more links, more of a way to use the existing number of links to establish the importance of Wikipedia in their public education work. I think what you mean is Linkypedia which was amazing but has been dead for years. There is a version of it on GitHub but this is way way beyond my level of technical expertise https://github.com/edsu/linkypedia. Would there be a way of installing it on WMFLabs or similar as a service? John Cummings (talk) 12:03, 2 June 2021 (UTC)
@John Cummings: The way I used to do this with LinkSearch was to up the page limit to 5000 in the URL then click Next Page until I hit the end. It wasn't a great method. Doing a Quarry query per the above is probably the best way to get a static number. We built a simple tool for longer term tracking for the library at Wikilink. It's not easily configurable by people who aren't us but if you have a need for some monitoring please feel free to drop me an email and we can set that up. Samwalton9 (WMF) (talk) 12:06, 2 June 2021 (UTC)
Thanks very much Samwalton9 (WMF), is there a plan to make this more open or people can make requests (with a public table like BaGLAMa 2? It would be super helpful for people working with GLAMs, I know several Wikipedians in Residence who would use it. Please could you run a query for UNESCO (http//:*.unesco.org) (they have different language versions) and EIB.org. Thanks again. John Cummings (talk) 13:58, 2 June 2021 (UTC)
Hi Xaosflux thanks so much, this is very helpful. One question for the UNESCO one specifically, unesco.org is structured by language the same way Wikipedia is, e.g en.unesco.org is the English version, fr.unesco.org is the French version etc. I see in the query you've structured the URL as https://org.unesco.%, I'm assuming this takes this into account? Is there a way urls should be structured so they take this kind of thing into account? Thanks again. John Cummings (talk) 16:03, 2 June 2021 (UTC)
@John Cummings: the % in there is capturing all of them, if you wanted it to be more specific you could put in "org.unesco.en%" for only en.unesco.org for example. — xaosfluxTalk16:24, 2 June 2021 (UTC)
Izno this is extremely helpful, thanks very much, one question, what is different between insource:unesco and insource:/unesco\.org/ ? When I replace UNESCO with WIPO in the search it seems to break and only show 4 results. Thanks again. John Cummings (talk) 16:16, 2 June 2021 (UTC)
@John Cummings: Without a link to your query I cannot troubleshoot why it broke. The first part starts the search by looking for any text "unesco" appearing somewhere in the wikisource. The second part is a regular expression that looks for the text unesco.org. The period is 'escaped' (because a period has another meaning in regex without escaping). I have also added an 'i' after the second search to make it case insensitive, since you seem to be dealing with some cases where the case might change. --Izno (talk) 16:28, 2 June 2021 (UTC)
Izno, thanks so much for the explanation, I'll write these up as instructions somewhere for other users who want to run similar searches. John Cummings (talk) 16:33, 2 June 2021 (UTC)
Text in non-Latin scripts should never be in italics or boldface. This is especially true for Chinese, which becomes unreadable in bold or italics even to many who could usually read it. Nobody deliberately typesets Chinese in bold or italics, but many infoboxes do so, as they automatically format text in italics or in bold (for example, book titles). Depending on the infobox, this can be fixed either with a language code parameter or in the morepedestrian way of using {{noitalic}} or {{nobold}}. The problem is how to actually find all these violations: is there a way to search for any text displayed in bold or italics that uses a given non-Latin character set? —Kusma (talk) 09:35, 30 May 2021 (UTC)
@Kusma:: Searching for all bold, Chinese text would probably require you to write a script to slowly scrape all of Wikipedia and process the resulting HTML. It does not sound fun! The reason I say this is that there are very many ways for text in an article to become bolded, and it is beyond our search engine to recognise all of them. The only feasible way to see what is bold is to render the page. However if you are particularly concerned about infoboxes then post at Module:Infobox suggesting that the relevant non-Latin scripts are detected with a Lua ustring pattern and made non-bold or non-italic automatically. User:GKFXtalk22:47, 1 June 2021 (UTC)
@GKFX: I am only concerned about templates. If infoboxes and citation and external link formatting templates no longer display bad bold or italics, almost everything is done. I'm not sure whether full automation as you suggest or a tracking category so other issues (language tagging) can be done is the best way forward, but I'll post at the relevant talk pages tomorrow or so. Thank you! —Kusma (talk) 22:58, 1 June 2021 (UTC)
It doesn't seem easy (main problem is I want to detect things that have NOT been tagged, but should be), so perhaps a scraper script is better after all? I would appreciate any creative ideas :) —Kusma (talk) 13:27, 3 June 2021 (UTC)
I forkedMdaniels5757's excellent markAdmins script, in spite of it being configurable, bc I wanted to use icons with tooltips instead of acronyms in paranteses separated by slashes. I managed to ungroup them in the xml from the user links, so it can be tooltipped, but haven't figured out how to do that. Any suggestions? I tried this. but evidently that's not it. I gather I need to create abbr nodes, but I'm fumbling in the dark. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎07:55, 1 June 2021 (UTC)
Well, if you're not talking about abbreviations (but icons), <abbr></abbr> is incorrect markup. Would you be happy with a <span></span> using a title attribute? I believe most browsers (I'm ignoring smart phones, as usual) show item titles as a tooltip when you hover. See Template:Tooltip (and its code) for ideas. — JohnFromPinckney (talk / edits)11:44, 1 June 2021 (UTC)
Yes, absolutely; my mistake. Yet, the issue is creating the separate span nodes, each with its titles. Basic js, I'd imagine. Cheers. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎04:33, 4 June 2021 (UTC)
Need Template Help if Possible - Tried to upgrade the "Template:Human timeline" to a somewhat larger (for better font sizes) template at => "Template:Human timeline/sandbox5" - some of the Note links (on the right side margin) seem OK (4 of 10), but several Note links (6 of 10) are NOT OK (no linking?) for some reason (except for these 6 Note links, all other of the 28 links in the template are OK) - I'm stumped with the 6 (NOT OK) Note links at the moment - any help with these particular links would be greatly appreciated - Thanks - Stay Safe and Healthy !! - Drbogdan (talk) 14:56, 4 June 2021 (UTC)
@Buidhe, Hike395, MarMi wiki, and Rummskartoffel: (and others) - Thank you for your comments - QUESTIONS: is there a Fix for this problem Display Bug? (click-and-drag-mouse doesn't seem to work) - or should this concern be presented (escalated?) to some other appropriate Wikipedia discussion group? - if so, which one exactly? - in any case - Stay Safe and Healthy !! - Drbogdan (talk) 21:31, 4 June 2021 (UTC)
Drbogdan Thanks for creating this new timeline at Template:Human timeline/sandbox5. I agree that it's an improvement on the original (for accessibility), but unfortunately it doesn't fix the linking issues for me. Hopefully someone else on this noticeboard will have an idea for fixing that. (t · c) buidhe21:52, 4 June 2021 (UTC)
Done@MarMi wiki and Rummskartoffel: (and others) - Possible Solution - all links now seem to work OK (using => WinTel10/ChromeBrowser/DellXPS8900) in "Template:Human timeline/sandbox5" - after removing/commenting-out => "annotations-width=8.8" in the template coding - hope this helps in some way - iac - Stay Safe and Healthy !! - Drbogdan (talk) 00:53, 5 June 2021 (UTC)
var, calc, background-image do not make it through the sanitizer I believe, and space-evenly isn't supported by all browsers. Between the two, the majority of the CSS in your module is invalid for use on Wikipedia.
Secondly, please return it to your sandbox. It is clear to me that is not something that we need on Wikipedia. Izno (talk) 05:08, 3 June 2021 (UTC)
And indeed, when I check my console, I see that much of the inline styles of the HTML has been stripped to <div style="/* insecure input */"></div>. There is a possibility your Lua could be amended to remove the offending inline styles and then you would get some output, but it would not necessarily be the output you want (one of the styles in question would be acceptable with TemplateStyles, but I leave that as an exercise to the reader) Izno (talk) 05:12, 3 June 2021 (UTC)
Apparently calc works, which was fun to learn. I assume you understand why url does not. var simply does not work and will not work at any point in the near future, so accept it. :^) Finally, see WP:TemplateStyles. Izno (talk) 07:29, 3 June 2021 (UTC)
@Izno: Just a little help once again. Can you please point out those properties which are causing the problem. Right now the properties containing var are the problem. I got that. Is there any more issue that you could point out. I also want to ask where can I see this output you saw: <div style="/* insecure input */"></div>Harsh Rathod Poke me!08:00, 5 June 2021 (UTC)
url() is insecure and is stripped from wikitext. It is allowed in TemplateStyles for a small set of websites. The output was in browser console. Izno (talk) 17:23, 5 June 2021 (UTC)
Edit to Template:Country data Czech Republic broke links
The protected Template:Country data Czech Republic was edited today to set shortname alias = Czechia. This breaks some uses of the template that uses the country name to generate a link to an article. For instance {{flb|CZE}} now expands to non-existing Czech Republic, while before it expanded to Czech Republic. The same for Template:fl-rt, Template:fl19, Template:flw and others. Can someone please fix this somehow? I cannot edit the protected template, nor am I sure what is the right course of an action (revert the change? rename the articles? create a redirect? add link alias-floorball exception? (though other uses apart from floorball might be affected too))
Sidenote: While I personally do not object the use of "Czechia", it seems that the change violates a consensus: Talk:Czech Republic#Czechia.
@Prikryl: Ignoring the 'broken' aspect of this which I will comment on the template talk page, I don't see the sidenote as relevant. Czechia redirects to Czech Republic, so it seems reasonable to support that in the template. Izno (talk) 16:07, 5 June 2021 (UTC)
@Izno: It seems that the consensus is that the "Czechia" shall not be used on Wikipedia (yet). Though, I'm not going to argue with you about that here. It's not the right place. It just seems that this should be properly discussed before making such major changes. --Prikryl (talk) 19:59, 5 June 2021 (UTC)
Wikipedia welcome
Occasionally, I am welcomed on a different-language Wikipedia (so far Italian and German). This seems to happen spontaneously. ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 14:40, 5 June 2021 (UTC)
This is acceptable if you actually made an edit on the wiki in question; it is understandable, if annoying, if you visit the wiki without making an edit. But a few days ago I got a notification from Tamil Wikisource, which I am absolutely 100% dead certain I had never visited until I followed that notification to find out what it was all about, only to find this notice, which I cannot read one bit. I'm pretty sure the WMF has decreed that welcome bots should not send messages to people who have never edited the wiki in question, but does the restriction also apply to humans? Courtesy notification to Sridhar G, who is not a bot. --Redrose64 🌹 (talk) 19:42, 5 June 2021 (UTC)
WMF has not decreed such a thing. That is entirely a (or several) particular community's decision.
There is a gadget out and about that can have you 'visit' every wiki such that you get all the welcomes out of the way at a single time. I don't have it to hand. Izno (talk) 20:01, 5 June 2021 (UTC)
I believe what happened with Tamil Wikisource is that edits you made to Module:Protection banner/config were imported to English Wikisource on August 20 (as part of an import of Template:Category disambiguation with the "include templates" box checked), and then imported from there to Tamil Wikisource on May 26 (as part of an import of s:Module:List with the "include templates" box checked), which automatically created local accounts for all users who had edited the page. The same thing happened to me via Module:Documentation (which was imported to English Wikisource as part of the same batch-import on August 20 and then imported to Tamil Wikisource on February 3 as part of an import of s:Template:Bad page scan), except for some reason I was never welcomed. * Pppery *it has begun...21:46, 5 June 2021 (UTC)
All users that have edited an page that was imported only get assigned if the admin clicks to assign the edits to global users and include all edits of the page. Both are off by default. That kind of action is deliberate. Notice the "w>" in the history of the en.wikisource module (Module:Documentation), (next to MusikAnimal) it indicates that that edit is not assigned to him.--Snævar (talk) 23:26, 5 June 2021 (UTC)
They are contributors to a page imported at that time.[7] None of them had ever visited the wiki. If they had then their accounts would have been created when they did instead. This is why my proposal at meta:Welcoming policy says: "A wiki is only allowed to post welcome messages to users if their account was originally created at the wiki, or the user has at least one non-imported edit there." PrimeHunter (talk) 22:02, 6 June 2021 (UTC)
A Redirect which goes to the correct Wikipedia article, but to the wrong place within the article
After a Teahouse discussion, the two people helping me there suggested that I ask here. (For a quick summary of that discussion, see the 6 June 2021 comment by User:Ganbaruby.)
Here are two examples (each going to a wrong location within the target Wikipedia article, COVID-19 pandemic in Illinois):
Anchor name specified: UIUC response to COVID-19 pandemic is a Redirect which does not end up anywhere near the anchor titled "UIUCresponse".
User:Ganbaruby reported that the Redirect he/she tested misbehaved in Safari, but worked correctly in Chrome. (All my testing was in Safari; sorry: I don't know how to determine what version of Safari I was using.) CWBoast (talk) 02:16, 7 June 2021 (UTC)
I think I can reproduce it (in Chrome). What happens for me if I click one of those links is that it first scrolls to the correct location. However, the huge bar chart with the number of cases ({{COVID-19 pandemic data/United States/Illinois medical cases chart}}) is still expanded. Then, at some point, it's collapsed by some javascript, and that reformats the text, which moves the target section up. --rchard2scout (talk) 07:33, 7 June 2021 (UTC)
This is expected with autocollapse and custom collapsibles as used by that chart. Its one of the reasons we tend to avoid using them. —TheDJ (talk • contribs) 10:25, 7 June 2021 (UTC)
This is not possible offhand, though occasionally requested. (Maybe someone has built or identified a tool on Toolforge since the last time?) Izno (talk) 12:41, 7 June 2021 (UTC)
You could try using WikiBlame, which might be able to find things that stayed in the article for a while before getting removed, but I honestly doubt it'd do a particularly good job. May be worth a shot, though. – Rummskartoffel (talk • contribs) 12:56, 7 June 2021 (UTC)
I actually started work on such a tool years back. I stopped due to difficulties with parsing the diff pages, but I've recently finished work on a different tool that might solve that problem. If I can get it working, I'll be sure to post a notice here and a few other places. ᛗᛁᛟᛚᚾᛁᚱPantsTell me all about it.13:21, 7 June 2021 (UTC)
@RoySmith: the which are similar component is going to be a far stretch. If you just want to search for a phrase or string, you could use Special:Export to export all of the revisions of the page, then just use a text search tool on the output file. (Limited to 1000 revisions). — xaosfluxTalk14:14, 7 June 2021 (UTC)
Infobox language family
In the infobox at Bantu languages, the Southern Bantu branch doesn't show up, even though it is present in the code. I found out that the number of displayable branches is capped at 20 (something I hadn't realised as of yet), and the parameter "child21=" results in an internal error message. What's the best way to deal with this awkward situation? --Florian Blaschke (talk) 19:16, 6 June 2021 (UTC)
You could have added more entries separated by <br> in child20. But the whole idea of displaying a list of numbered parameters seemed pointless so I replaced it with one long parameter containing the list.[8] Maybe the infobox documentation should suggest this. The list is rather long for an infobox but it's not my field. PrimeHunter (talk) 21:51, 6 June 2021 (UTC)
Thank you! A quick workaround I thought of afterwards was simply deleting the first entry, which isn't even a real branch but a link to the geographical classification by Guthrie. But your solution works too and has the flexibility I was looking for. I only just remembered the similar solution employed in Tibeto-Burman languages. --Florian Blaschke (talk) 15:01, 7 June 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 8 June. It will be on non-Wikipedia wikis and some Wikipedias from 9 June. It will be on all wikis from 10 June (calendar).
Future changes
The Wikimedia movement uses Phabricator for technical tasks. This is where we collect technical suggestions, bugs and what developers are working on. The company behind Phabricator will stop working on it. This will not change anything for the Wikimedia movement now. It could lead to changes in the future. [9][10][11]
Searching on Wikipedia will find more results in some languages. This is mainly true for when those who search do not use the correct diacritics because they are not seen as necessary in that language. For example searching for Bedusz doesn't find Będusz on German Wikipedia. The character ę isn't used in German so many would write e instead. This will work better in the future in some languages. [12]
The CSRF token parameters in the action API were changed in 2014. The old parameters from before 2014 will stop working soon. This can affect bots, gadgets and user scripts that still use the old parameters. [13][14]
Can I configure my Wikipedia account to use MFA? Ideally, it would work like GitHub or NPM where I can use an authenticator mobile app and have some recovery codes which I can use in case when my phone is stolen. Grillofrances (talk) 02:25, 7 June 2021 (UTC)
Thanks so it seems that only some groups of users have 2FA available so I need to request such access. IMO each user should be able to configure a 2FA and for the most privileged users, it should be even required. Grillofrances (talk) 23:13, 7 June 2021 (UTC)
We know. Though I don't agree with the opinion, there is a large opinion that WMF 2FA is too hard to deal with.
That said, the WMF-reason for not having it for all people today is that is not easy to fix if you lose your account password and your scratch codes and particularly requires human resources that have higher-paying jobs than to deal with such a user. Izno (talk) 01:30, 8 June 2021 (UTC)
Keeping track of new pictures in a wide category tree
Hello,
I'm a marine biologist specialized in Echinodermata. I would like to be informed of any new picture of these animals so I can review the identification and, when useful, add them to the relevant Wikipedia articles. But as there are over 7000 species of the, of course I can't check all the categories every day. I used to benefit from Ogrebot's newsfeed for a long and useful time but it is no longer working. Do you guys know any other way I could get such uploading newsfeed ? (knowing that I'm not a hacker).
This is basically the same question as the "Best way to monitor Category:Requests for unblock", so the same answers apply.--Snævar (talk) 09:20, 8 June 2021 (UTC)
I'm not sure I understand what you mean, Snævar. Xaosflux, it's for pictures that are not yet on commons but will be added in the future. I want to know when people add them, so I can review and use them. Thanks ! FredD (talk) 11:38, 8 June 2021 (UTC)
Problem with major intersection in an article about a road
Take a look at Dillon County under "Major intersections" in South Carolina Highway 38. There are two places called Oak Grove, South Carolina. The one that had an article before I added the one in Dillon County has more people and is obviously more notable.— Vchimpanzee • talk • contributions • 17:43, 6 June 2021 (UTC)
I am making a tool for monitoring Category:Requests for unblock. I looked at the IRC feed but it does not seem to monitor additions and removals from categories.
Right now I am using the API to ask for its members every few minutes, but it would be nicer if I could get a feed as I prefer not to bother the API too much. Does anyone know a better way to monitor for additions and removals from a category? HighInBCNeed help? Just ask.02:44, 8 June 2021 (UTC)
@HighInBC: so if you are "making a tool", can't your tool just query the category? The other option mentioned above involves the watchlist. One thing you could try would be to make another account, add those categories to that account's watchlist - then use the watchlist RSS feed for that account to see the changes. — xaosfluxTalk10:38, 8 June 2021 (UTC)
I am querying the category right now. Interesting idea about the watchlist. Right now it is not using and account as it is read only. But that is a good idea. Thank you. HighInBCNeed help? Just ask.10:48, 8 June 2021 (UTC)
Yes that is perfect. I can supply the API that token, and the timestamp of my last check and get a full update in just one call. I no longer have to store the previous state to compare against to find out what has changed. Thank you, and thanks to everyone else who gave advice. HighInBCNeed help? Just ask.11:15, 8 June 2021 (UTC)
@Xaosflux: But as you can see in the URL, isn't a feed essentially just an API call returned in Atom XML rather than JSON? I can't imagine it's less server-expensive than simpler API calls that don't require tokens. Nardog (talk) 13:43, 8 June 2021 (UTC)
@HighInBC: Oh, you're making a tool! Sorry I was off. I may be still preaching to the choir, but: AFAIK there are two basic ways: list=recentchanges can give you the specific edits that caused additions/removals, but it's limited to last 30 days and doesn't detect changes by way of transclusions; list=categorymembers doesn't have those limitations but only gives you additions, not removals, and only timestamps, not revisions. Nardog (talk) 12:08, 8 June 2021 (UTC)
@HighInBC: The Wikimedia EventStream API shows both category additions and removals. It's a stream (hence makes your tool actually work in real-time, rather than every 10 minutes or so), and doesn't require any authentication or atom XML parsing. – SD0001 (talk) 03:32, 9 June 2021 (UTC)
Thank you, I was looking for a real-time solution instead of polling. This is very useful. I have learned a lot from this thread. HighInBCNeed help? Just ask.04:04, 9 June 2021 (UTC)
Is there a way to build a URL which gets you to the "Change protection" screen, but with fields pre-filled? You can do that with block user, via Special:Block, i.e. this, but as far as I can tell from the wikimedia manual, there's no version of that for page protection. What I'm hoping to do is be able to have a script which creates a link you can click on to take you to a pre-filled out form, which you just have to review and click the "Confirm" button. I know I can do it directly through the API, but if I can do it with a clickable link, that would be preferable. -- RoySmith(talk)14:00, 8 June 2021 (UTC)
@RoySmith The URL params should match the name attribute of the input fields. You can inspect the DOM to see what they are. This technique should work on most all forms in MediaWiki. Here's an example that touches almost every field: [15]. — MusikAnimaltalk16:24, 8 June 2021 (UTC)
Five minutes ago, my Watch list looked like it always did. Is it my imagination, or are the square green boxes on the left new (as of the last 5 minutes, it seems). Some are green. Some are the same old blue color they always were. I went over and looked at my Commons watchlist, and the little boxes to the left are all the same blue color. I'm not sure what the green color is supposed to signify. Clicking on any color of them, does not change the color. So, what is the purpose? — Maile (talk) 23:08, 9 June 2021 (UTC)
You may have accidentally enabled "Display green collapsible arrows and green bullets for changed pages in your watchlist, page history and recent changes (unavailable with the improved Watchlist user interface)" in the Preferences. Nardog (talk) 23:15, 9 June 2021 (UTC)
Yeah, I saw on Preferences that it wasn't new. But something on mine made it suddenliy really stand out, to the point of being annoying. Perhaps a font or some other something updated on my system to cause it. My browsers - Firefox and Chrome - both showed it the same way. — Maile (talk) 23:35, 9 June 2021 (UTC)
@Redrose64 and Nardog: I think - perhaps - why this suddenly stood out at me. And maybe this changed. I've been looking at my watch list on Commons and Wikisource. They both have the green and blue square boxes, but look normal to me. The difference - both of them display the time of edit to the right of the edited page name. The one here on Wikipedia displays it to the left of the page name, between the boxes and the article name, creating what seems like a wider gap between the boxes and the rest of it. Or maybe it's my imagination, but that seems to be the difference to me. — Maile (talk) 00:52, 10 June 2021 (UTC)
Is the global Lua function type effectively overrideable?
So I made a function in Module:Lua class (last function) that tries to override/build upon the default type to support new "types". I was testing it in User:Alexiscoutinho/sandbox 1 and noticed that it behaves differently if you are previewing. When you view the sandbox normally, it should contain "table table table". However, if you preview the page (without edits of course) it would now contain "class Base instance" (which is desired). What is going on? Why is the type function behaving differently? Alexiscoutinho (talk) 21:33, 10 June 2021 (UTC)
Thank you very much! I completely forgot how to fully purge a page. Furthermore, I thought purging was only client side, thus I got quite surprised when the issue persisted in an anonymous tab. Alexiscoutinho (talk) 01:13, 11 June 2021 (UTC)
@Qwerfjkl: What do you mean the editing window is appearing in the text editor, what steps do you perform for this to happen? addmylinks has some buttons that look the same as the ones in QuickEdit, but that's the addmylinks text editor and not QuickEdit.
Do you mean when you save a page using QuickEdit, the page appears in the "My Links" section on the sidebar? – BrandonXLF (talk)19:15, 3 June 2021 (UTC)
At first, pages appear normally, but when I click the 'quick edit' button, the 'My links' section is replaced with the normal 'QuickEdit' editor. (I'm using the Vector skin, but it might be affected by User:TheDJ/mobileVector.css.) ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 06:59, 4 June 2021 (UTC)
When I load any .js page on the desktop version of Wikipedia, I cannot paste or copy text, so I have to switch to the mobile version. Is there a way to get a round this? (I think WikEd allowed me to view text normally.) ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 16:52, 6 June 2021 (UTC)
There's certainly something odd about the edit box in some circumstances. If you don't get an edit cursor when you click in the edit box, click in the edit summary box instead, then use ⇧ Shift+Tab ↹ to get to the main edit box. You should then be able to click with the mouse on the desired editing point. --Redrose64 🌹 (talk) 18:22, 6 June 2021 (UTC)
Can you copy and paste when you log out? Safemode does not load any user js page. What do you mean by "load any .js page" in "When I load any .js page on the desktop version of Wikipedia, I cannot paste or copy text"? Are there circumstances where you can copy and paste in desktop? PrimeHunter (talk) 09:17, 10 June 2021 (UTC)
@PrimeHunter Normally, when editing on a mobile device in desktop mode (or mobile mode), I can double- or triple-tap on a word to highlight, and an option to copy or paste appears above the selected text. However, in .js pages, this doesn't happen; the text selection behaves wierdly, and instead of options appearing above the selected text, a '...' Box appears, with multiple options that don't work. ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 21:28, 10 June 2021 (UTC)
Oh, you are talking about editing js pages. Loading a js page usualy means a load command to run the script in the page. Try clicking the <> icon at the left of the toolbar to switch between the normal editor and the code editor. PrimeHunter (talk) 21:37, 10 June 2021 (UTC)
Is there any way to make links to other language wikipedias, commons, Wiktionary, etc, open in a new tab? I sometimes forget that they don't behave like normal external links and it's annoying. DuncanHill (talk) 00:59, 11 June 2021 (UTC)
Weird, I didn't know that was possible (nor do I understand *why* that is possible) but none of the examples before that were external. Mike Peel (talk) 12:47, 11 June 2021 (UTC)
Traditionally most of the other websites that had special interwiki prefixes were other wikis or other free content sites, with some other sites like Google thrown in for convenience. The main difference between the interwiki map links and normal external links (other than the styling) is that interwiki map links do not have the "nofollow" attribute. (I once got so annoyed at Wikia/Fandom's advertising that I worked a bit to turn "nofollow" on for links that go there). —Kusma (talk) 13:04, 11 June 2021 (UTC)
@SD0001: Thanks, that certainly works for commons, wikiquote, etc, doesn't work for the interlanguage links down the left-hand side. @Jonesey95: and @TheDJ: actually right-click and open in new tab is what I use, but as I mentioned I sometimes forget, and @Mike Peel: they may or may not be external but they behave like external links in that the preferences I set here do not apply there, and things like templates that work here do not work there. Anyway, thanks all. DuncanHill (talk) 16:30, 11 June 2021 (UTC)
Does anyone know what's happening here? It seems like an attempt to create a redirect to a centralized talk page, but the "../" isn't something I've seen before and it's not currently working. {{u|Sdkb}}talk17:58, 7 June 2021 (UTC)
@Sdkb: This has always worked fine for me, and I use it sometimes. Here's a demo, just follow the breadcrumbs. Do any of these not work for you? I'm on Win 10/Vivaldi, but I checked it on iOS 14.2 mobile as well. Haven't incorporated it into a redirect, though, so maybe there's something squirrely there. Mathglot (talk) 01:56, 12 June 2021 (UTC)
Hidden maintenance category shows up at top of main category list
Why am I seeing one hidden category showing up at the top of the main category list at the bottom of an article?
On the article Simone de Beauvoir, the first category I see is "Wikipedia articles with WORLDCATID identifiers", a maintenance category. The next two cats are Simone de Beauvoir, followed by 1908 births; these two correspond to the first two categories explicitly in the article wikicode. The first category is not in the wikicode anywhere, which is not surprising, as it is a maintenance category that is probably dragged in by a citation with an |oclc= param.
@Drbogdan, unless someone wants to count the number of grains of sand on Mars, I don't see why the image dimensions wouldn't be downscaled by 4, and the image saved as jpg, like they did here https://photojournal.jpl.nasa.gov/jpeg/PIA24663.jpg, file size 15 MB (link found here). Those who really need the 2.4 billion pixel image can always go back to NASA. Ponor (talk) 14:19, 11 June 2021 (UTC)
Have you ever been frustrated because there's no way to map namespace numbers to namespace names in a database query? I hereby present Quarry 55924 for your amusement. There's hacks, there's ugly hacks, and then there's hacks that are so ugly you're proud of them. -- RoySmith(talk)16:40, 12 June 2021 (UTC)
You don't even need to join pagelinks for that: quarry:query/55928.I had a somewhat similar idea a while ago, but using a sandbox with links to Talk:Talk, User:User, User talk:User talk, and so on, so I could just use pl_namespace and pl_title out of pagelinks, and not have to worry about some joker breaking the query by creating extra subpages or inbound links or whatever. I couldn't figure out how to get it working with the main namespace, though.quarry:query/55930/User:Cryptic/ns, to make it concrete. Munging the titles with MID() etc. like you did could do it, but it stops being brief enough that I'd feel ok pasting it into the middle of real queries. Still better than what I usually resort to like in e.g. quarry:query/50255. —Cryptic20:04, 12 June 2021 (UTC)
My go to is including the namespace number and then using regex change it to {{subst:ns:N}} and then putting it in a sandbox and then it works. A better system would certainly be nice though. --Trialpears (talk) 20:22, 12 June 2021 (UTC)
Namespaces aren't consistent across wikis - see the identical list rendered on 5 wikis - English, Commons, WikiData, Meta, and German.
Since Quarry allows querying different wikis just by changing the specified database from enwiki_p, the SQL should also (preferably) be wiki-neutral to maximise reusability. The method in Quarry 55915
dresses the namespace/page-title as a wikilink in a wiki-neutral way (skipping the use of regex Trialpears mentioned). Thankfully namespace 14 is consistently Category (or equivalent) in needing the colon prefix for all wikis AFAIK. The resultset can be taken as a wikitable & sandboxed on the corresponding wiki.
I've learned from experience that the file namespace (consistently 6) also should have a leading colon, otherwise that looks great and I will use it next time. --Trialpears (talk) 08:15, 13 June 2021 (UTC)
@Cabayi: Namespaces -2, -1, and 0 through 15 inclusive are consistent in their purposes across all WMF wikis, and I think across all MediaWiki wikis also. The non-negative numbers also always occur in pairs: all subject namespaces have an even number, and its corresponding talk namespace is always one greater. Whilst the local names do vary (e.g. User: might be Utilisateur: (French) Benutzer: (German), etc.), the name used on English Wikipedia may be used for all of them (e.g. try following fr:User:Redrose64), with one pair of exceptions. These exceptions are namespaces 4/5, which have been localised here to Wikipedia:/Wikipedia talk:, and as you noticed at Commons, Meta etc. the local names are Commons:/Commons talk:, Meta:/Meta talk:, etc. The default non-localised name for these two namespaces is Project:/Project talk:, and they work everywhere - try following Project:Namespace, meta:Project:Namespaces or indeed ru:Project:Пространства имён.
For namespaces numbered 16 or above, not all of them exist on all wikis, but for those that use a given number, its purpose is the same for all wikis that use the same number - for instance, Module:/Module talk:, where they exist, are always 828/829 although as with namespaces 15 and below, the local names may vary.
@Trialpears: The File: namespace is like the Category: namespace in that an initial colon is necessary to make a link instead of to activate a feature. --Redrose64 🌹 (talk) 08:31, 13 June 2021 (UTC)
Since this method doesn't touch on the actual namespace names (localised or global) no adjustment is needed on that front. The SQL is still wiki-agnostic.
0mtwb9gd5wx, thanks for improving the documentation! Usually we keep documentation on a separate documentation page such as Template:DistroWatch/doc if it is longer than a line or two which, I have now done. You also accidentally added it in a way that the documentation was included when the template was used on pages which is another reason it's much simpler to have a dedicated documentation page. --Trialpears (talk) 08:54, 13 June 2021 (UTC)
0mtwb9gd5wx, I think WP:UNCONF applies and I'd decline it. Other renamers may view it differently and perform the rename. Once done, other users may also see it as WP:UNCONF and report you to WP:UAA.
It also has the potential to break stuff. Many templates handle usernames and strip leading namespaces. I doubt any of them have been tested for this level of misdirection.
About a month ago I brought up the possibility of removing the education program talk namespace at Wikipedia:Village pump (technical)/Archive 189#Education program namespace removal. This was prompted by T217137 and similar tasks which resulted in removal from other Wikis following that the namespace was emptied either through deletion or moves. Currently there are 0 pages in the Education program namespace and 1,427 pages in the Education program talk namespace. My plan is to move most pages from their current location of Education program talk:FOO to Wikipedia talk:Education program archive/FOO. Some pages that have no significant content (such as the supported by wikied messages) will/have been deleted outright without archiving. Assuming there are no objections in the next few days I will proceed with this plan. --Trialpears (talk) 08:38, 13 June 2021 (UTC)
I've now deleted that and 75 other similar grading subpages after verifying that they indeed doesn't have any possibly useful content. --Trialpears (talk) 16:36, 13 June 2021 (UTC)
FYI, during the last Community Wishlist Survey, this task received a great support and should be taken over by The Community Tech team in the coming months. Pamputt (talk) 21:44, 13 June 2021 (UTC)
Cats visible in mobile view for mobile users
Hello, probably just my device, or, are there any other mobile users experiencing categories being visible(in a block like manner) whilst in mobile view? is there a d.i.y manner for me to remedy this? Celestina007 (talk) 20:34, 10 June 2021 (UTC)
WP:ITSTHURSDAY, I think. This appears to be intentional and working as designed, as step one of a change in mobile category display. See T246049, which appears to have changed the mobile view to display categories. It may be followed by T152199, which intends to make them collapsible. This change appears to have been announced as part of the often inscrutable list of updates linked from the above Tech News (and always posted well after the Tech News, so you have to remember to check the link sometime after the Tech News posting). See mw:MediaWiki 1.37/wmf.9#MinervaNeue for the description of this change. I am not always great at following the phab/gerrit/WMF bread crumbs, and never great at understanding why they work on this fiddly stuff instead of fixing real bugs, so I could have any of this wrong. – Jonesey95 (talk) 20:54, 10 June 2021 (UTC)
This change removes the old component that used to do categories on mobile, which had a half dozen open bugs associated with it. So... yes, it fixed real bugs. Izno (talk) 21:09, 10 June 2021 (UTC)
Jonesey95, Rummskartoffel, Izno thanks, I think I’d probably just opt to hide it (if it doesn’t translate to it being invisible in desktop view also) You coders/script creators and all template editors in general are gifted. This aspect of editing just overwhelms me. Celestina007 (talk) 21:19, 10 June 2021 (UTC)
Interesting question. The native API result suggests usersearch.py is more severely broken; it finds 0 edits, instead of the many the API finds in the last 500 revisions. Also a bug? (It is "part of the English Wikipedia", xaosflux, in the sense that the tool is linked to from every page like that where I said I started, though at toolforge.org. So? I can follow up on their talk if @Σ: doesn't respond to the ping and/or others may help. E.g. if I'm confused and these aren't bugs.) --50.201.195.170 (talk) 02:12, 9 June 2021 (UTC)
50, we link to many external tools - but we (the editors and admins of wikipedia) don't maintain them. If it is broken to the point being useless, we can remove the external link from the interface or change it to something else - but we can't actually do anything about it being broken any more then if the RIR links on the bottom of your talk page had an error - we have no access to the codebase for that tool, only its maintainer does. Removal or replacement can be discussed at the associated message here: MediaWiki talk:Histlegend. — xaosfluxTalk10:26, 9 June 2021 (UTC)
The people that maintain them often watch this page, or know what the issue is, how to fix it, or provide alternatives. I think this is a perfectly acceptable venue. Because I saw this discussion, I've changed MediaWiki:Histlegend to point to what should be a fully working tool (disclosure, authored by yours truly). — MusikAnimaltalk22:57, 10 June 2021 (UTC)
Did anyone find the 2 bugs reproducible (or not reproducible)? I guess so, but I don't see anyone mentioning one way or the other. --50.201.195.170 (talk) 23:14, 9 June 2021 (UTC)
Up until recently there was not a way to search for edits by IP addresses. At some point this became possible, but I have simply never gotten around to implementing this. →Σσς. (Sigma)10:00, 11 June 2021 (UTC)
IP addresses have actor IDs just like accounts, so if you go by that it should be the same query for both IPs and accounts. If you want to query for IP ranges (which XTools only very recently added support for, but it's been technically possible since late 2017), then that does require a different query. Is the source code published anywhere? I'm happy to help. — MusikAnimaltalk17:21, 11 June 2021 (UTC)
Ha! So not broken, after all. Though you could strip out the blank spaces server-side, I suppose. Edge case, for sure! Anyway in light of this I've changed MediaWiki:Histlegend to point back to Sigma's. — MusikAnimaltalk05:22, 12 June 2021 (UTC)
Thank you for helping, MusikAnimal! Wrong?: Σ: the link above doesn't work because there's a space at the end Umm...gaslighting? The links in my OP above obviously don't have spaces in 'em. They can't; they're bare URLs. (And no %20's either.) No, they were certainly not working and work now. --50.201.195.170 (talk) 00:05, 13 June 2021 (UTC)
Clicking on "Find edits by user" today is taking me to a new user interface at xtools.wmflabs.org. It was a different one until yesterday. Is there any notice or discussion around this? How can I remove the pie charts that show up in that page? I could not find a way to see a customized result, even after logging in to that site. Or, how can I go back to the older interface? - JayTalk13:02, 11 June 2021 (UTC)
I had seen it earlier, and did not know it is related to my query. How is it related to my query? What has happened? - JayTalk14:13, 11 June 2021 (UTC)
Ok, so the earlier tool was called WikiBlame and it is broken, and xtools is a temporary replacement? And this discussion is the place where we can get updates on this? - JayTalk14:30, 11 June 2021 (UTC)
On second thoughts, can we retain XTools permanently? It is superfast as compared to the earlier tool, and I can live with the pie charts if not customizable. Or is there a way a user can customize the tool to be used for "Find edits by user" via Preferences? - Jay(Talk)05:31, 12 June 2021 (UTC)
Won't there be thousands of users using the "Find edits by user" functionality, who will be affected? Shouldn't there be a vote, or a customize option? - Jay(Talk)20:13, 13 June 2021 (UTC)
I think it'd make most sense to just have both, similar to how it's done with "Find addition/removal" (e.g. Find edits by user (Alternate)). – Rummskartoffel20:29, 13 June 2021 (UTC)
As I said, I went the proper route of soliciting feedback before and no one engaged. It definitely doesn't seem worthy of a formal RfC. There's no way to customize this on a per-user basis, so Rummskartoffel's suggestion sounds best. I've restored Sigma's tool as the long-standing default and made XTools an alternate. Hopefully that satisfies everyone :) — MusikAnimaltalk23:39, 13 June 2021 (UTC)
Uncaught ReferenceError: capitalScript is not defined
at <anonymous>:1:701
at domEval (load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:11)
at runScript (load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:13)
at enqueue (load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:11)
at execute (load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:14)
at doPropagation (load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:6)
Is there a command which, for lack of a better term, I'd call {{scrunch}} which would do what I requested here? (what someone on the help desk calls "Yes the gap is kind of annoying") Pi314m (talk) 00:58, 14 June 2021 (UTC)
Please look at Linden General Hospital. The History section begins without a gap, following the LEAD. By contrast, in Glens Falls Hospital there is a long gap, what help desk calls "kind of annoying" (and which, via either an existing flow control command, or a new one, I'd like to reduce, if not eliminate). Pi314m (talk) 09:25, 14 June 2021 (UTC)
@Pi314m: I'm not seeing any abnormal gaps in Glens Falls Hospital. Are you referring to the space that is being used by the table of contents section (which is not on the first article because it is so short)? This is normal layout, as seen in this random article I just loaded: Gigan. How this appears may vary with the screen width of the reader, and also will look a lot different on mobile browsers. — xaosfluxTalk10:26, 14 June 2021 (UTC)
Yes, I think what you are seeing is the normal space created by the table of contents (TOC). To see a comparable space in some of our highest-quality articles, take a look at Babe Ruth, or Four Freedoms (Rockwell), or Ring ouzel. Articles without a TOC (i.e. with three or fewer headings,) will not have this gap. – Jonesey95 (talk) 16:26, 14 June 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
Logged-in users on the mobile web can choose to use the advanced mobile mode. They now see categories in a similar way as users on desktop do. This means that some gadgets that have just been for desktop users could work for users of the mobile site too. If your wiki has such gadgets you could decide to turn them on for the mobile site too. Some gadgets probably need to be fixed to look good on mobile. [16]
German Wikipedia, English Wikivoyage and 29 smaller wikis will be read-only for a few minutes on 22 June. This is planned between 5:00 and 5:30 UTC. [18]
All wikis will be read-only for a few minutes in the week of 28 June. More information will be published in Tech News later. It will also be posted on individual wikis in the coming weeks. [19][20]
googling Special:CreateAccount for example, or googling Special:UserLogin or any other special page to my knowledge, I've seen it always says "No information is available for this page. Learn why." Why is that? Also some others like googling Wikipedia:Sockpuppet investigations also don't seem to work either, but it never seems to work for special pages. Why does it not work specifically for special pages, or for any other specific page? It doesn't appear to be the same as noindexing, as you can search for the page, it just doesn't give you info. 54nd60x (talk) 13:44, 12 June 2021 (UTC)
The page at one time was robots.txt blocked to avoid it being indexed. Then google started indexing pages even if they showed up in robots.txt and said we should be using noindex. But noindex cannot easily be detected if the page is also in robots.txt (unless the google bot finds some per chance traversal path that has the pages linked from elsewhere)... and those Special pages were never removed from the robots.txt. Similar for SPI. —TheDJ (talk • contribs) 14:42, 12 June 2021 (UTC)
@TheDJ: Why were special pages avoided being indexed? And does it mean that the magic word __NOINDEX__ would have no use as it can't be added to a special page. If the pages are not noindexed in robots.txt, why doesn't it give you info for special pages? And in what circumstances would it be best to avoid searching for special pages? 54nd60x (talk) 01:24, 13 June 2021 (UTC)
@54nd60x: Special pages can't be indexed because there is nothing to index: a special page doesn't actually exist until you visit it. When you follow a link to a page like Special:Preferences the page is generated specifically for you; once served to you it's deleted again; and what you receive is different from what other people receive when they follow exactly the same link. --Redrose64 🌹 (talk) 07:39, 13 June 2021 (UTC)
@Redrose64:Special pages can't be indexed because there is nothing to index: a special page doesn't actually exist until you visit it. Could you please explain that? Does it mean that special pages have visible content but they aren't actually editable as it's just transclusion of MediaWiki pages? Also, why is Wikipedia:Sockpuppet investigations and perhaps some other non-special pages noindexed? Does it have to do with privacy? 54nd60x (talk) 10:41, 13 June 2021 (UTC)
Special pages could be indexed but they aren't. It's a choice that we ask external search engines to not index special pages. They place noindex in their html, and http://en.wiki.x.io/robots.txt says Disallow: /wiki/Special:. Special pages are generated on demand and often rely on url parameters. Many special pages often change content, depend on the user viewing them, or have rather arbitrary content we don't want random searchers to find instead of our articles. PrimeHunter (talk) 21:41, 13 June 2021 (UTC)
@PrimeHunter: I was browsing special pages and found that the content can be changed with ?uselang=$1, as in Special:SpecialPages vs. Special:SpecialPages?uselang=es. However, I noticed that the content of Special:MathWikibase is always written in the wiki's language (i.e. it's in Spanish on es.wp and in English on en.wp but the contents are not messages and the content of the page cannot be changed based on user preferences.) Where is the source content of these special pages that allows it to be different across different wikis, and how can it be edited if some major change must be made? 54nd60x (talk) 00:48, 14 June 2021 (UTC)
OK - so as should have been surmised from above, this is not something we can fix directly here on the English Wikipedia - WP:BUG requests can be opened about any special page that is being presented by an extension. In the case of Special:MathWikibase for example, I've opened: phab:T284960. — xaosfluxTalk23:26, 14 June 2021 (UTC)
There is something broken at Schengen Area: the text seem to be mostly in red with some huge font size. I have slow connection to the internet at the moment and can not debug it. Can someone look at it? --Jarekt (talk) 09:57, 14 June 2021 (UTC)
@Firefly: I've seen a number of similar notes about infobox populations ([21][22][23]), but couldn't see the issue myself. Would this be the same issue? CMD (talk)
Chipmunkdavis - yes, almost certainly, as these articles also use the template mentioned above. Now that I've fixed it I can't see any oddities on those pages either. firefly ( t · c ) 16:34, 14 June 2021 (UTC)
My apologies, my intention was to update the figures to latest UN estimates (which are now outdated). I've now fixed the proposed change in the sandbox, and tested it in the aforementioned pages, with no hiccups. If no one finds any other problems with it, I'll put it into production in a few days. Cheers. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎00:05, 15 June 2021 (UTC)
Template/Module for penultimate revision
Anyone seen a template or Lua function that returns the link or id to the last revision of a given page before the current one? (Bonus points for one that returns the last revision by a given user!)
@Redrose64 There are non-biographies that have BLP material, but those do not have such editintros. I am wondering if it is possible for any article with the {{BLP}} tag on the talk page to have the BLP editintro show up. Aasim (talk) 08:18, 15 June 2021 (UTC)
Here's the code:
if(mw.config.get('wgNamespaceNumber')===0){$(function(){if(document.getElementById('disambigbox')){addEditIntro('Template:Disambig_editintro');}});$(function(){varcats=mw.config.get('wgCategories');if(!cats){return;}if($.inArray('Living people',cats)!==-1||$.inArray('Possibly living people',cats)!==-1){addEditIntro('Template:BLP_editintro');}});}
is that really needed though ? Kinda a slippery slope if you ask me and before we know it we have even more banners that no one reads on every single page. —TheDJ (talk • contribs) 09:06, 15 June 2021 (UTC)
Test range for block warning?
From a sub-discussion at WP:VPW#IP Masking Update, you get the attached warning box when you try to block a sensitive IP range. What produces that? Is there some edit filter? Something in the interface? Back-end wikicode? And, more to the point, how can we add 192.0.2.0/24, to give admins a safe (sandbox) range to experiment with? -- RoySmith(talk)17:08, 15 June 2021 (UTC)
Quick CSS question, what's the default left margin between the edge of an image and the article body text? I'm trying to get a table to match. {{u|Sdkb}}talk23:34, 14 June 2021 (UTC)
Ah, so they don't. In any case, do not assume the margins you have in your skin will reflect the other skins, esp. mobile. Izno (talk) 02:57, 15 June 2021 (UTC)
In any case, your browser console will tell you the widths and heights of an arbitrary block on a page, and tell you why those are. --Izno (talk) 02:59, 15 June 2021 (UTC)
yeah table floatleft and float right are exceptions and use different margins historically. I do wonder why though... Maybe that is something that should be discussed ? —TheDJ (talk • contribs) 09:09, 15 June 2021 (UTC)
@TheDJ and @Izno, I'm guessing it should be discussed, yeah; tables should automatically have the same sort of margins as pictures, without me or anyone else needing to dig deep into the CSS to figure out what they should be and set them manually. I didn't have any luck using my browser console to find the margin number, so I still don't know what it should be, nor where I'd go to propose a more general fix. {{u|Sdkb}}talk20:15, 15 June 2021 (UTC)
Besides that, a quick lintercheck suggests there are syntax problems in the if (scriptmanger == 0) { section on line 42. — xaosfluxTalk18:49, 14 June 2021 (UTC)
Also, there's a typo; you declare scriptmanager, but check scriptmanger, without the second "a". scriptmanger is presumably undefined, and will throw an error. Writ Keeper⚇♔18:54, 14 June 2021 (UTC)
Works for me (and by that I mean that it doesn't return an error, I don't know if it actually works). In your common.js you're missing ending bracket "}" for it (the same you have at the end of the above code example). MarMi wiki (talk) 21:15, 15 June 2021 (UTC)
When newcomers had the Reply tool and tried to post on a talk page, they were more successful at posting a comment. (Source)
Earlier this year, the Editing team ran a large study of the Reply Tool. The main goal was to find out whether the Reply Tool helped newer editors communicate on wiki. The second goal was to see whether the comments that newer editors made using the tool needed to be reverted more frequently than comments newer editors made with the existing wikitext page editor.
The key results were:
Newer editors who had automatic ("default on") access to the Reply tool were more likely to post a comment on a talk page.
The comments that newer editors made with the Reply Tool were also less likely to be reverted than the comments that newer editors made with page editing.
These results give the Editing team confidence that the tool is helpful.
Looking ahead
The team is planning to make the Reply tool available to everyone as an opt-out preference in the coming months. This has already happened at the Arabic, Czech, and Hungarian Wikipedias.
@Elizium23: It is not clear if the power outage was planned by the Internet Archive or an electric utility. They have been offline for more than 12 hours already. Are there any other details regarding this? LSGH (talk) (contributions) 03:11, 16 June 2021 (UTC)
Request: Semi-automated tool for identifying/fixing name typos in references
I'm not sure if there's an appropriate place to suggest this apart from here, so just throwing it out: I just corrected "Rieter" to "Reiter" in a ref, something that happens fairly often. I think if we made a semi-automated tool that found automatic last names in references (the same way the citation tool does) and spit out a list of instances where it differed by only one letter, it'd get quite a few hits. Does this sound feasible, and if so, would anyone like to take it on? {{u|Sdkb}}talk05:04, 16 June 2021 (UTC)
Need someone with a screenreader to test 2 options
What is your understanding of screen readers that you think the left margin is going to be evident? Has it still not occurred to you that the emphasis on the visual is leading you to focus on inappropriate solutions? — JohnFromPinckney (talk / edits)23:44, 15 June 2021 (UTC)
I have no idea what you are talking about. I am not emphasizing the visual. That's obvious, since I am asking for people with screenreaders to check out the 2 sandbox options, and tell us if their screenreader notices anything more than the indented left margin on one of the options. If the screenreader notices that at all. --Timeshifter (talk) 23:55, 15 June 2021 (UTC)
The question is a bit contrary to your intent: by saying "Other than the left margin..." you're implying screen readers will behave differently based on the left margin size. (The left margin makes no difference to the underlying document model and won't affect screen readers. Incorrect HTML will affect the document model, and it can cause unexpected behaviour.) But in the larger scheme of things, the tables don't really comprise a list. If they did, they should be marked up that way even if not being displayed side-by-side. It would be better not to put the tables into a list solely to take advantage of the default padding of list items. isaacl (talk) 00:22, 16 June 2021 (UTC)
I am concerned about accessibility problems. And simplifying things. If it causes no accessibility problems, then simple is better. User Prime Hunter proposed an even simpler solution with templates. So that is a 2nd option. The left margin is a problem when the 2 side by side tables are in a narrow screen and the tables wrap (one drops below the other). A horizontal scroll shows up sooner when there is a left margin. --Timeshifter (talk) 00:35, 16 June 2021 (UTC)
On a side note, PrimeHunter is just suggesting to hide the details of additional wrapper HTML inside templates. However with the style property applied directly to the table, no wrapper HTML is needed. isaacl (talk) 00:47, 16 June 2021 (UTC)
Isaacl. See Help:Table#Side by side tables. "style=display:inline-table can not be added within the table wikitext, because long captions will mess things up in mobile portrait view, or other narrow screens. It must be added outside the table wikitext."
I put both pages in tabs in the same window and switched back and forth as I changed the window width. I don't see any differences. (Even if there were differences, it doesn't make semantic sense to put the tables into a list solely to take advantage of the styling of list items.) isaacl (talk) 01:51, 16 June 2021 (UTC)
Also, isaacl, someone who uses a screen reader (Graham87) says that the div example works fine. See talk diff. Here is the div example below. I pasted it here so that others can check it out too.
Total number of matches played in official competitions only.
How come there are still issues with the font size when using the classic vector skin gadget (e. g. as to the rendering of references and categories)?--Hildeoc (talk) 14:08, 15 June 2021 (UTC)
@AKlapper (WMF): See "Preferences → Appearance → Vector classic typography (use only sans-serif in Vector skin)". Browser: Firefox (current). I just checked once again, and the problem is that with that gadget enabled, the font size of regular text on any English Wikipedia page is simply tiny.--Hildeoc (talk) 18:08, 15 June 2021 (UTC)
Did someone change the text in the orange box that comes up when someone leaves you a new talk page message, and if so, could we change it back (or for me to change my stylesheet to change it for myself)? It's longer now and "Talk" is randomly capitalised. Anarchyte (talk) 06:47, 16 June 2021 (UTC)
It's capitalized because "You have a new Talk page message" is shown instead of the normal "Talk" link. It does look odd but I wouldn't call it random and I'm neutral on the capitalization. PrimeHunter (talk) 10:04, 16 June 2021 (UTC)
Yes, I see the same issue. The use of two partial Infobox templates, that have to be used as a pair, has got to be A Really Bad Idea (and is probably to blame) — GhostInTheMachinetalk to me11:51, 16 June 2021 (UTC)
This is one of those quirks of the MediaWiki parser - tests for blank/non-blank occur before HTML comments are stripped out, so the presence of such a comment counts as non-blank. --Redrose64 🌹 (talk) 12:31, 17 June 2021 (UTC)
HTTP 502
Is there any reason why Wikipedia appeared to be inaccessible around one hour ago? All other websites were working properly, but every time I tried to go to any Wikipedia page, a WMF error page showed up instead. LSGH (talk) (contributions) 10:19, 15 June 2021 (UTC)
@The wub: Thanks. The incident report is not yet available, but someone identified the cause to be a router issue. How is it that only Wikipedia was affected while other WMF sites were not affected? LSGH (talk) (contributions) 03:11, 16 June 2021 (UTC)
One possible explanation is that you were not logged in to those other sites and saw cached pages, while the connectivity issue between the edge datacenter and rest of the network prevented non-cached requests (including all logged-in requests) from being properly served by MediaWiki itself. Majavah (talk!) 13:01, 17 June 2021 (UTC)
RfC location on validity of generated HTML
I am locked in a weird argument with another editor about whether Wikipedia should serve valid HTML or not, and what qualities wikitext needs to/does not need to/should not have. We're so entrenched in our opinions that I think I'm going to make my first RfC. My question (placed here, since HTML generation is kind of a technical matter) is: where would such an RfC be most appropriate? Help talk:HTML in wikitext? Somewhere else? — JohnFromPinckney (talk / edits)19:21, 15 June 2021 (UTC)
@JohnFromPinckney: so - it depends on what you want to "change". If you want a new technical functionality added, metawiki/phab/mailing lists would be a better place. If you want to change the manual of style about what type of markup is preferred then somewhere related to MOS may be best. Your premise of whether Wikipedia should serve valid HTML though sounds like a software bug (i.e. if we are "serving" invalid html in our rendered pages to viewers) - if the bug is coming from something we've done on-wiki - then place it near where that it, if the bug is coming from the linter/rendering engine/etc - then open a bug report. Fixing technical malfunctions doesn't normally require as much of a consensus gathering exercise like changing the manual of style would. One thing you've done well is to think about this ahead of time - and really what page hosts the RfC isn't super important, so long as it is advertised at any of the venues that are applicable and ends up being well-attended. — xaosfluxTalk20:48, 15 June 2021 (UTC)
Thanks, xaosflux. Probably before an RfC (now that I think more about it) I should just find a good place to get some discussion going. I'm not sure I want to change anything; I'd like to know if I'm the only person here who thinks the HTML we emit should be valid or not. If I am, then I'll have to decide how long to tilt at that windmill. If I'm not alone, I'd like to see where it's codified as policy/guideline/MoS item/whatever. And if it's not codified, but the community finds it important, then I'll try to get something formal.
And I'm not talking about a software bug; my colleague thinks <li> elements (which they add manually into the wikitext) don't need <ul> or <ol> so insists on removing them or reverting my additions. I say the resultant page is invalid HTML (although my browser seems to display it nicely) and so is wrong for us to do (and recommend, for example, on Help talk:Table). — JohnFromPinckney (talk / edits)21:34, 15 June 2021 (UTC)
A bare <li> that occurs outside of any enclosing <ol>...</ol> or <ul>...</ul> pair is indeed invalid. See HTML 5.1 spec, the part beginning "Contexts in which this element can be used". --Redrose64 🌹 (talk) 22:02, 15 June 2021 (UTC)
Yes, Redrose64, I am aware that it's invalid, the argumentative editor doesn't care. And GhostInTheMachine, I agree, there's very little need here, but that doesn't stop our colleague from preferring it. The difficulty I'm having is convincing them that there is value in emitting valid HTML. In the discussion started on their talk page, they are demanding some evidence that screen readers have problems with broken HTML, or that weird code causes accessibility problems. Quote from that editor "I see none". :-( — JohnFromPinckney (talk / edits)22:41, 15 June 2021 (UTC)
JohnFromPinckney, What do you mean by "valid HTML"? At a minimum, I think everybody would agree that it needs to be well formed, and as far as I can tell, it meets that. Even when you try to intentionally break things like this (view the source to see what I mean).
Talk pages are known to be aberrant on the point, so it's not an interesting comparison. John is specifically (and correctly) identifying that the <li> element in question must be contained inside either a <ul> or <ol> element, which is worth worrying about in the mainspace. Izno (talk) 22:57, 15 June 2021 (UTC)
As I understand it, the issue is the side-by-side tables example at Help:Table. I copied the example to User:Isaacl/table example to isolate it. The MediaWiki software doesn't try to fix it up, and it remains malformed HTML in the generated output. isaacl (talk) 23:40, 15 June 2021 (UTC)
Yes, but the reason it's not strictly necessary is the the Mediawiki parser will automatically close the <p> for you, such that the page output does not contain invalid HTML code. (See here for an example.) It doesn't do that for uncontained <li>, so that's not a relevant comparison here. Writ Keeper⚇♔00:38, 16 June 2021 (UTC)
@Writ Keeper: Even if the Mediawiki parser did not automatically close the <p>, the page output would not be invalid HTML. If you refer to the HTML 5.1 spec it has a whole paragraph headed "Tag omission in text/html" which is quite comprehensive - there are very few situations where the omission of a </p> would not be valid. In XHTML it was quite true that omitting a </p> tag was invalid (in fact, all closing tags were mandatory), but the MediaWiki software was switched from outputting XHTML to HTML5 in September 2012. --Redrose64 🌹 (talk) 07:57, 16 June 2021 (UTC)
@Timeshifter: That w3schools page says nothing of the kind - the five letters "valid" are absent, and the </p> tag is mentioned eight times - all of which are in examples, it is not used in text. The <p> tag is found ten times - eight of which are in examples, paired with the aforementioned </p> tags, one is in a summary label and just one is in the text. There is nothing that explicitly states that <p> without </p> is also malformed HTML or any variant on that. --Redrose64 🌹 (talk) 07:57, 16 June 2021 (UTC)
I know w3schools regularly comes up at the top of a search results list, but I do not consider them authoritative. My go-to place is the W3C, to which there is absolutely no connection. — JohnFromPinckney (talk / edits)10:37, 16 June 2021 (UTC)
I personally prefer and recommend MDN. I have found it to be very reliable, and it specifically provides accessibility advice for many topics. – Rummskartoffel11:39, 16 June 2021 (UTC)
@JohnFromPinckney: I happened to find this XfD discussion just now, while tracking down an entirely unrelated issue. Strange coincidence. On the topic of where to go for HTML and CSS questions, I usually use W3Schools because of their easy readability. But, yeah, when I really want to dig into the details, W3C is certainly the authoritative reference. -- RoySmith(talk)13:32, 17 June 2021 (UTC)
For some use cases it is possible to emulate the ifexist effect with css, by using the selectors a.new (to select links to unexisting pages) or a:not(.new) (to select links to existing pages).
Manipulating the base link color is generally not user-friendly UI (that we ignore it for sports is a sports issue... we just shoot for basic color contrast accessibility). So... why would we want to make something like this? Izno (talk) 00:38, 17 June 2021 (UTC)
Still pondering on whether or not this type of presentation is a good idea or not, but in the meantime, the tables where you've used it should have a legend to explain the meaning of the symbol. (The documentation of the template also needs more detail.) isaacl (talk) 14:21, 17 June 2021 (UTC)
Please see Template talk:SockBlock § Inaccessible on mobile. I'm posting this here to get more eyes on the issue, and hopefully formulate an appropriate edit request. In short, {{SockBlock}} cannot be viewed on mobile unless the desktop website is used, or by opening into the edit window for source material. Unfortunately, I don't have the technical know-how to be able to convert the template properly. Ideas welcome. Sdrqaz (talk) 17:27, 17 June 2021 (UTC)
Thanks, Xaosflux; I did suspect that it was due to {{tmbox}}. Given that the Phabricator task is "low"-priority and doesn't really capture the problem (the task is for those which are configured differently instead of being invisible altogether), would it be possible on our end to make a workaround (like {{uw-sockblock}})? This template is used as a block notice, after all. Sdrqaz (talk) 18:35, 17 June 2021 (UTC)
I'm sorry but are you sure? That talk page has 62 watchers and this one has over 3,000. Moreover, it seems to be an issue inherent to {{tmbox}}; moving away from it for {{sockblock}} seems the best solution. Sdrqaz (talk) 19:11, 17 June 2021 (UTC)
At the very least, could we update out documentation so all of our message boxes clearly say whether they are visible on mobile or not? A quick ctrl-f mobile gave me zero information about this. For things that we want to be visible on mobile, we should use code that makes them visible on mobile. —Kusma (talk) 19:25, 17 June 2021 (UTC)
(I was told to ask here from the Teahouse, see Wikipedia:Teahouse#Possibility_of_viewing_tooltips_on_mobile.) Is it possible to view tooltips in articles (the ones indicated by dotted underlining) while viewing Wikipedia on mobile? Although the underlining appears, when that is touched the tooltip doesn't appear. This is the same even when using Desktop view in mobile. NS-Merni (talk) 06:15, 16 June 2021 (UTC)
When you edit a module, the second infobox says "Editors can experiment in this modules's sandbox (edit | diff) and testcases (edit) pages." when it should be "Editors can experiment in this module's sandbox (edit | diff) and testcases (edit) pages.". Alexiscoutinho (talk) 20:14, 17 June 2021 (UTC)
How come when inserting a copied Wiki link to a lemma with special characters like http://en.wiki.x.io/wiki/%C3%9Cbermensch (for the article Übermensch) and selecting To a wiki page in the "Insert link" form of the 2010 wikitext editor toolbar, the output is still a faulty link %C3%9Cbermensch with a warning saying The requested page title contains invalid characters: "%C3". – as has been the case for many years now? This seems to happen in all Wikis, by the way. (Has there ever existed a corresponding Phab ticket or anything similar dealing with this issue?)--Hildeoc (talk) 21:29, 16 June 2021 (UTC)
@Hildeoc: this sounds like GIGO. Why would try to insert an external link to an internal article in to an article? This seems to work fine if you just insert the article name. — xaosfluxTalk13:41, 17 June 2021 (UTC)
@Xaosflux: It's actually somewhat easier to just paste the URL into the form, and have it automatically transformed into a Wiki link – isn't that exactly what that function is supposed to be used for …?--Hildeoc (talk) 14:10, 17 June 2021 (UTC)
No, if I understand you here is what you are doing:
Pasting a full URL in to the INSERT LINK function
Changing the selector to "To a wiki page" (despite not actually putting in a wiki page)
Seeing the error
Now, notice that if you don't but a percent-encoded URL in that step 1 the next steps end up being:
Get a warning that "Page does not exit"
End up inserting a bad string such as [[http://en.wiki.x.io/wiki/Apple]]
So even if it wasn't for the refusal to accept your percent encoded url here, the output is still undesirable, you didn't put in a "wiki page" so your output is not a "wiki page". — xaosfluxTalk14:32, 17 June 2021 (UTC)
@Xaosflux: But with URLs linking to Wiki lemmas without special characters the conversion into Wiki links works perfectly fine (except for the automatic addition of underscores for spaces, which don't affect the validity of the converted links created, though), and to me that's a very useful thing. Notice that the form automatically shows the following message when you paste a Wikipedia URL: The URL you specified looks like it was intended as a link to another wiki page. Do you want to make it an internal link?--Hildeoc (talk) 14:41, 17 June 2021 (UTC)
@Hildeoc: OK, that will require a software change to mw:Extension:WikiEditor. I've opened a feature request here: phab:T285144. Feel free to subscribe to the FR, correct anything I put wrong on it, or add to it. Please note, there are currently 97 requests in the backlog for that extension, so it could be a long time for someone to work on this. — xaosfluxTalk13:14, 18 June 2021 (UTC)
Dear xaosflux, how can I thank you? This is really nice of you – thank you so much for taking the time and trouble to take care of that! I've just created a Phab account and subscribed to your request. I do hope it will be considered at some point. For the time being, I wish you all the best--Hildeoc (talk) 13:27, 18 June 2021 (UTC)
Accounting of deleted (legit) templates in contributions
I've seen a couple of similar discussions, but they are very old. Every time I talk about deleted contributions below I mean DEPRECATED TEMPLATES.
Is showing private (only user who did it can see) deleted contributions still a concern to the point of not being worth moving this discussion forward?
Does the "number of edits" counter in Preferences count deleted contributions?
Is it acceptable to undelete a deprecated template, to blank and redirect it, only so that its edit history is preserved (and contribs)?
(1) Yes, keep in mind almost all deletions are entire pages that have many contributors. (2) yes. (3) it really depends, "deprecated" alone isn't usually a reason to delete something - deletions are carried out under the deletion policy, most deletions should be logged as to the reason it was deleted and there are various reasons to undelete something. You can read all about that in the Wikipedia:Deletion policy and the guidance at Wikipedia:Requests for undeletion. (Bonus #4) Don't worry about accounting for contribution "edit counts" - all sorts of things can impact it, and at over 500 edits the exact figures don't really matter for anything important. See Wikipedia:Edit count and the links on that page for way too much on that topic. — xaosfluxTalk15:49, 18 June 2021 (UTC)
It generally has been the policy to delete unused templates, even if previously used. It does break old versions of pages, though, so I wish there was a bit of a better way. But leaving template namespace filled with things long-ago deprecated really isn't ideal. Elli (talk | contribs) 15:51, 18 June 2021 (UTC)
@Elli: I think it would be nice if Wikipedia had a separate old/historical branch where all the old stuff lives. This would keep the namespaces clean and wouldn't break old revisions and people's hearts. Alexiscoutinho (talk) 20:10, 18 June 2021 (UTC)
I tried archiving a discussion at WP:ANI with OneClickArchiver and it's archiving the wrong one. The automated edit summary is labelling the discussion that I want to archive, though, so it seems like some sort of heading detection issue. Anyone know what the problem is? (please mention me on reply) —Tenryuu 🐲 ( 💬 • 📝 ) 22:36, 18 June 2021 (UTC)
Actually, it may have worked previously and enabled users to add code to the bottom of a JavaScript page by leaving the heading field empty. PrimeHunter (talk) 22:48, 18 June 2021 (UTC)
There is a discussion going on at wikiproject basketball regarding presentation of statistical data for players in basketball articles. I won't reprise all of the discussion, but some editors prefer only ratios for certain statistics. While I support the inclusion of ratios, there are times I think it's helpful to see the statistics as totals per year. While showing both might be an obvious answer, the inclusion of all fields in a single table generates an unwieldy sized table.
I'd like to explore an option to eat our cake and have it too.
I'm intrigued by the presentation in many info boxes about locations that gives you the option of showing one of several maps or all maps with a simple click of a radio button. As an example, see the info box for Alliance,_Ohio, which allows you to choose whether you want to present a map of Ohio, a map of the United States, a map of North America, or all three. While I can see that this is done with a switch, I haven't figured out whether this same concept could be applied to tables.
In short, I'm trying to figure out if I can set up something so that a single table (probably defaulting to the ratios) is presented but the user has option to switch to the totals table or to see both.
I've presented example data in User:Sphilbrick/Stats_options. The first table labeled "ALL" is not what I want as it is too wide, although it does reflect the standard presentation in typical schools media guides or record books. What I would like to do is have both tables available, the one labeled "Ratios" and the one labeled "Totals", with one showing as a default and the ability with a radio burtton to switch to the other or see both simultaneously. I am concerned about Wikipedia:Manual of Style/Accessibility issues. My hope is that the broad acceptance of the map option means the accessibility concern doesn't apply. (As a related point, I had some initial concerns about the header row showing popup abbreviations, but I see MOS explicitly allows that MOS:NOHOVER.)
Can someone tell me if this is feasible and if so how to implement it?
As a secondary technical question, I generally like to use the visual editor to add tables by constructing them in Excel and then copying and pasting them. I haven't figured out how to handle the heading row using visual editor. Is there something I'm missing?--S Philbrick(Talk)14:11, 17 June 2021 (UTC)
Pppery, Looks promising, User:Sphilbrick/sandbox but I made it work by putting the individual tables and a separate page and transcoding them. This would work as a one-off but not workable for hundreds of articles so I need to figure out how to include the data in the template. S Philbrick(Talk)14:37, 17 June 2021 (UTC)
I thought I'd try including the two tables in the article in a special section that would be hidden, then transclude the two tables into the switcher template but that attempt failed miserably. I'm guessing that it is not acceptable to do a tranclusion on a page that incorporates information from that same page; I got a template loop error.
I did find an approach that works, see User:Sphilbrick/Stats example, but I'm worried that this will be frowned upon because I dropped the tables into sub pages then transluded.
Using subpages for permanent content that is meant to be part of the encyclopedia.
I think the rationale is that a reader should never have to visit a sub page to see relevant content but that doesn't apply in this case given the transclusion. It does mean that updates to information by editors would require visiting there but editors unlike readers are not going to have any difficulty locating the data.
I put together a simple example so that it's visible here rather than having to go to the user page to see it:
The only downside is that it requires me to construct the tables manually, line by line and using the sub pages allows me to copy and paste tables from Excel, but I have a feeling the community is not going to accept sub pages. Any thoughts? Thanks for your help, one of these two options should work.--S Philbrick(Talk)11:22, 19 June 2021 (UTC)
File at ENWP
How I can see original upload//log/page for this image (it was moved to Commons)? Current file has comment "I created this myself 20/4/08" by Taopman. What could a comment from those years mean? I found source which says that take a picture was only possible on 27 April 2007. Eurohunter (talk) 12:24, 19 June 2021 (UTC)
The file was moved, which makes it a bit harder to find. Original log is here: [26]. The original license information was {{self|cc-by-3.0}}. There is nothing else in the deleted history. —Kusma (talk) 12:45, 19 June 2021 (UTC)
Discussion about using CSS margins to modify kerning
Reaching out a second time, since there is little participation (only 4 editors so far) for this wide-reaching and highly-visible change. Discussion has now turned to whether the currently-live implementation (applying a 0.2em margin around parenthesis for certain letter-based cases in the Nihongo template) should be reverted, kept, or modified. — Goszei (talk) 01:18, 20 June 2021 (UTC)
Harv cite issues, Harv errors abound...
Harv cites are not my jam but when Harv errors and warnings litter an article it annoys me so I try to fix them. That being said, since they are not my jam, they are often a source of frustration to me because I usually cannot easily figure out exactly what is wrong. So, O Wiki Technical/Coding Mavens Who Understand All The Niceties of Harvard/Sfnm/etc Citations... could someone PLEASE take a look at Chester A. Arthur and tell me why it is littered with 44 "Harv errors"? I tried to fix one - the Abbot cite - but was unsuccessful. If you could explain it here, maybe step-by-step in somewhat plain English so my addled brain can understand that would be awesome. If you don't mind, please don't fix the issues at the article - explain it here and I'll get at it myself. I know it's probably something incredibly simple but I am just not seeing what the issue is at the moment. Thanks. Shearonink (talk) 17:24, 19 June 2021 (UTC)
With the version of the script you have installed, it correctly identifies that items in the Further reading section that you are citing with a Harv template should be in a separate section above the Further reading section. Perhaps "Works cited" or "Bibliography". The point of a further reading section is to contain works that you did not cite but have included for 'further' reading. Izno (talk) 17:41, 19 June 2021 (UTC)
Like Izno said, that's because you have references in a further reading section, instead of a bibliography section (or similar). Headbomb {t · c · p · b}17:46, 19 June 2021 (UTC)
Just for the record, I did not set the references up and I have had little to do with the article until (maybe) today. Chester A. Arthur is a featured article and the present referencing layout is how I found it. Btw, there is already a References section but it only has 2 Refname references + a single Sfn reference sitting in it... . Shearonink (talk) 17:59, 19 June 2021 (UTC)
{{Excerpt}} Gives the impression that I can transclude a specific table if it has an ID.
tables — Tables to transclude. By default all tables are transcluded. Same syntax as when transcluding paragraphs, but also: tables=Stats2020 — Transclude the table with id 'Stats2020'
Template:Campaignbox is not visible on mobiles and tablets even if desktop mode is turned on
Resolved
– Chrome "desktop view" is not Wikipedia's "Desktop" mode; the former does not request additional content such as navboxes, the later does (and works). — xaosfluxTalk07:52, 19 June 2021 (UTC)
When I look at War of the Fifth Coalition on my Android mobile with Chrome Version 91.0.4472.101, I cannot see the result of the template "Campaignbox Napoleonic Wars" and I get no warning that something cannot be displayed. Even if I turn on desktop mode I cannot see any result of any campaignbox. The same is true for my Android tablet with Amazon Silk. The campaignboxes are very important to allow the user to navigate easily from article to article. Ruedi33a (talk) 14:54, 18 June 2021 (UTC)
War of the Fifth Coalition in desktop mode according to Chrome on my mobile___=en.m.wiki.x.io/wiki/War_of_the_Fifth_Coalition, no campaign box
War of the Fifth Coalition not in desktop mode according to Chrome on my mobile=en.m.wiki.x.io/wiki/War_of_the_Fifth_Coalition, no campaign box
War of the Fifth Coalition url manually edited_____________________________=en.wiki.x.io/wiki/War_of_the_Fifth_Coalition, campainbox is visible but you cannot read it as the letters are too small unless you enlarge this part of the screen: visible but unusable for a simple user Ruedi33a (talk) 18:24, 18 June 2021 (UTC)
It works for me at [27] (link from following the Desktop view). Can you verify you successfully made it to the desktop mode (that your url is not en.m.wikipedia...). — xaosfluxTalk15:39, 18 June 2021 (UTC)
You shouldn't just use the browser's toggle if you're currently viewing the mobile version of a page. You should use the toggle at the bottom of the page itself ("Desktop"), which serves you the desktop version no matter what the browser's preferences are. Nardog (talk) 20:52, 18 June 2021 (UTC)
Works for me on desktop, too. Would it be possible to make it to where such templates appear on mobile w/o going to desktop mode? It seems that those templates are likely useful even for mobile viewers. Hog FarmTalk15:46, 18 June 2021 (UTC)
The fundamental problem with mobile seems that Wikipedians like to manually write things that work on the desktop with no regard for mobile. Developers then just code something automatic (like hiding certain boxes, or switching certain orders around, or the automated suggestions that replace some navboxes) to work around some problems with what Wikipedians have created. It might be possible to harness the power of Wikipedians to improve the mobile experience (Wikipedia's success is to use smart human labour instead of algorithms), but I wouldn't even know where to find the documentation. —Kusma (talk) 16:08, 18 June 2021 (UTC)
Not only that, navboxes also take up an exorbitant amount of the data of page delivery, especially on the more popular articles. Downloading that on a phone is not fun —TheDJ (talk • contribs) 17:48, 18 June 2021 (UTC)
Navboxes originated in the time before iPhones and the subsequent dominance of mobile Internet surfing. They just never made the jump from a design perspective. And also the size. I was going to play around with at least fixing the display part on MediaWiki wiki after navbox there got (fully) TemplateStyled, but TheDJ stopped working my edit requests ;). Izno (talk) 17:56, 18 June 2021 (UTC)
I'm not saying we should display standard navboxes on mobile (I'm not sure we should display them on desktop tbh), I'm trying to say we should design human-curated navigation tools that work well on mobile. —Kusma (talk) 20:12, 18 June 2021 (UTC)
Maybe we should have a new class navbox-mobile which is similar to navbox but renders on mobile. {{Navbox}} and other templates could have an option for which class to use. Editors would probably disagree which navboxes are important enough to show on mobile so we would get one more thing to fight over. PrimeHunter (talk) 22:56, 18 June 2021 (UTC)
I was unclear. I meant for navbox-mobile to render on both desktop and mobile, not forking to only render on mobile. PrimeHunter (talk) 09:50, 19 June 2021 (UTC)
No, the point I'm making is that must require a fork, because it's 'navbox' that causes these not to display on mobile. And it's not "display: none" CSS, it's "MobileFrontend rips the HTML out entirely". Izno (talk) 16:48, 19 June 2021 (UTC)
No, my idea would not require a fork. I'm suggesting that {{Navbox}} and friends get a new optional parameter, e.g. |mobile=yes. If the parameter is set then they add a new class navbox-mobileinstead of adding navbox. navbox-mobile should have the same styling as navbox so it renders the same on desktop, but it also renders on mobile since navbox is no longer present. This assumes the mobile developers don't decide to also omit navbox-mobile from mobile. We can make a guideline about only using |mobile=yes on small or essential navboxes. Individual navbox templates may pass on the parameter, e.g. using {{Academy Award Best Actor}} on Academy Award for Best Actor to render on that article in mobile, but not on the biographies. PrimeHunter (talk) 20:59, 19 June 2021 (UTC)
PrimeHunter, would that not require forking the navbox style into two classes though which would then need to be kept updated in parallel? I assumed that is what Izno meant. firefly ( t · c ) 21:08, 19 June 2021 (UTC)
I assumed Izno meant forking a lot of navbox templates. Defining an extra CSS class should be simple whether or not we call it forking. PrimeHunter (talk) 23:51, 19 June 2021 (UTC)
@Kusma:The fundamental problem with mobile seems that Wikipedians like to manually write things that work on the desktop with no regard for mobile. This is simply not true. The {{campaignbox}} template has been using class="navbox" for many years, and did so long before the devs started writing a special skin for mobile displays and provided the https://en.m.wiki.x.io/ domain to go with it. It was the devs who made a deliberate decision that any element belonging to the navbox class should not be displayed in mobile. So don't blame Wikipedians. --Redrose64 🌹 (talk) 07:19, 19 June 2021 (UTC)
@Redrose64: Sorry, I think my point did not come across. I was trying really hard to AGF with the devs and may have overshot. I'm just saying, there are good reasons why the devs chose a few years ago not to display navboxes. That they replaced them by RelatedPages still strikes me as bizarre (and somewhat insulting, I don't like being replaced by algorithms). The question is what we should do now. The most obvious solution to me is having separate navboxes, which according to @Iznomakes no sense. The next obvious solution is to cull and cut down desktop navboxes until they make sense also on mobile, and to display them everywhere again. —Kusma (talk) 08:34, 19 June 2021 (UTC)
(As to "forking", I agree that asking desktop editors to edit something separate that is only visible on mobile isn't going to work. But maybe there's something else we can do). —Kusma (talk) 09:52, 19 June 2021 (UTC)
A social solution won't work. It's been a decade and editors are still happily making very large navboxes. :\ Izno (talk) 16:51, 19 June 2021 (UTC)
It was a good decision not to display navboxes on mobile, they spread and grow like a cancer; in a few years every article will have a list of nearly all wikipedia articles in it. We already have infoboxes that are 5 screens long on mobile. I mean, who goes to wikipedia to *read* all these identifiers here? Ponor (talk) 12:50, 20 June 2021 (UTC)
FWIW, I've always like the navboxes and have found them extremely useful both as a reader and an editor to find related topics, and I use them all the time. As for the mobile version, I have no need for it even on mobile devices, and improvements in mobile phone technology that seemed to demand a mobile version in the first instance when they became capable of handling web pages now are making it obsolete. Hawkeye7(discuss)00:20, 21 June 2021 (UTC)
First I have to thank everybody joining this discussion about my question. I started as an ignorant human being and now I am on the long, cumbersome way to become a Wikipedian.
Summary: As a simple user I did not know the link "Desktop" hidden at the bottom of every wiki page. I have tried it now once and the result on my mobile was disastrous: the normal text is OK, the text of the infoboxes and campaignboxes is too small to be readable and must be increased manually. I will never use desktop mode again. If a wikipedian will say to me use "Desktop" mode to see the campaignboxes, I will know that this is an answer out of The Hitchhiker's Guide to the Galaxy: absolutely correct but does not help at all. The situation is that a user sees only 95-99% of this kind of wiki article on his mobile/tablet and gets no warning.
I suggest this solution: The simple infobox is a perfect replacement to solve the problem but the infobox seems to be too boring and too ugly for wikipedians. Compare Template:French invasion of russia mobile with Template:Campaignbox Napoleon's invasion of Russia. I suggest to replace each campaignbox with an infobox or something better than infobox that is visible on a mobile/tablet. Ruedi33a (talk) 11:00, 19 June 2021 (UTC)
We already have a meta-template for this, it's {{sidebar}}. That template also does not display on mobile, basically for the same reasons as the navbox above. I do not think this solution will do anything, and it's likely to be deleted locally anyway due to being a fork or converted to use sidebar, which will subsequently remove it from mobile. "I want to see these on mobile" isn't good enough to solve the problem. Izno (talk) 16:54, 19 June 2021 (UTC)
@FDW777: well this is the tech page, so not sure what you are looking for. I don't think we should be misusing infobox classes to house a navbox though. — xaosfluxTalk12:07, 20 June 2021 (UTC)
Template:Infobox military conflict contains a campaignbox parameter that allows campaign boxes to be included. Some campaigns are important enough to have many articles and some are not. But not all articles using a particular campaignbox is a campaign; only the top level one. Hawkeye7(discuss)00:29, 21 June 2021 (UTC)
It appears to mean any autoconfirmed user making an edit to a Pending Changes-protected page that doesn't have any pending edits (and thus their edit being automatically approved) * Pppery *it has begun...00:16, 21 June 2021 (UTC)
Firstly, Im not sure if this is the correct forum for "how to..." questions - if not, please re-direct me!
I have created a template related to the civil war and mercenary involvement in the Congo in the 1960's - this is the template, let's call it Target Template:
The section titled "Main Operations and Battles" contains a list of events that is described in another template - below is the one concerned - lets call this one Source Template:
The "Source Template" data for "Main Operations and Battles" is the primary data and is maintained in "Source Template" by unknown editors. To keep "Target Template" up to date, I can (a.) copy all the "Main Operations and Battles" wiki-links from "Source Template"' to "Target Template" [this is what I have done] and check from time to time that any changes made to "Source Template" is manually replicated to "Target Template;" or (b.) nest the data from "Source Template" directly inside "Target Template" so that it is only maintained in "Source Template."
Doesn't seem to work for templates - one has to use the "# tag" function instead..... I think we will have to maintain the data in two places! Please see attempt at Source TemplateFarawayman (talk) 23:29, 17 June 2021 (UTC)
There appears to be an undocumented limitation: Labeled section transclusion doesn't work when the marked section is part of a template parameter. PrimeHunter (talk) 01:53, 18 June 2021 (UTC)
Yes, there is a limitation when transcluding from templates that I never quite understood. A workaround is to use lua to to get the section content. I set up something that works in {{section/sandbox}} a couple of years ago but there was no follow-up. I've used it to show that it works in this case in {{Congo Conflict: 1960–1968/sandbox}}. There are a number of other modules (e.g. Module:Excerpt) that might be suitable and more developed. — Jts1882 | talk08:17, 18 June 2021 (UTC)
Is there a gadget to expand all collapsed sections on a page, including nested ones? An article like My Little Television needs over 50 clicks to show all its content, which can be inconvenient when searching. I could probably throw some JavaScript together to fiddle with the mw-collapsible-* classes but the idea seems too obvious to be original, so I expect this wheel has already been invented. Certes (talk) 17:00, 20 June 2021 (UTC)
$('.mw-collapsed').each(function(){$(this).data('mwCollapsible').expand()}); would do it, but they shouldn't be collapsed in the first place per MOS:DONTHIDE. Nardog (talk) 17:09, 20 June 2021 (UTC)
I, being a mobile user, cannot copy text from a page's normal view, and have to into editing mode to do it, which can be problematic on pages using {{PAGENAME}} and similar. Is there a way around this? ―Qwerfjkl | 𝕋𝔸𝕃𝕂 (please use {{reply to|Qwerfjkl}} on reply) 15:05, 21 June 2021 (UTC)
Maybe try a different web browser or app? I have no trouble copying text in the Read mode on mobile view (using Safari on iOS or Firefox on Mac OS). – Jonesey95 (talk) 15:22, 21 June 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Recent changes
The otrs-member group name is now vrt-permissions. This could affect abuse filters. [28]
Problems
You will be able to read but not edit German Wikipedia, English Wikivoyage and 29 smaller wikis for a few minutes on 22 June. This is planned between 5:00 and 5:30 UTC. [29]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 22 June. It will be on non-Wikipedia wikis and some Wikipedias from 23 June. It will be on all wikis from 24 June (calendar).
I'm looking for a template to strip formatting from a wikilink and give me the raw target. {{Delink}} works quite well but for piped links it gives me the label instead of the target.
It would seem that you should be able to modify delinkWikilink(s) at Module:Delink#L-38 to do that because titlearea (poorly named, that) gets the target article name so you could return it after Module:Delink#L-59?
If anyone can help, there is a gif I'm using on Cai Lun's article. However, at FAC, two users have expressed that it goes rather quickly between images (an insight I agree with); I attempted to use this website to slow down the process to five seconds per image, but when I uploaded it, it became exact some speed. Any advice or solutions would be appreciated—and apologies if this is the wrong forum for such a query. Best - Aza24 (talk) 22:38, 22 June 2021 (UTC)
Indeed! I seem to have been nothing but impatient, it probably took a second for it to update is all. Aza24 (talk) 00:20, 23 June 2021 (UTC)
Tooltip for WikiMiniAtlas coordinates inside a File template is cut off
When WikiMiniAtlas coordinates are inside a File template in an article, the tooltip is cut off by the image box and is not shown outside of it, as can be seen on this page. I tested this on Windows 10 in Firefox, Chrome and Edge. Does anyone else have this problem? DxhaFFer (talk) 14:37, 22 June 2021 (UTC)
Is there any way to get a gallery to display on a specified number of lines (for desktop)? I got the one here to display nicely on one full line on my display, but when I tried a different monitor it went onto a second line. There doesn't seem to be anything at Help:Gallery tag, so I'm thinking it might be necessary to use some HTML in the widths parameter or something like that. {{u|Sdkb}}talk23:08, 22 June 2021 (UTC)
Why don't you want me (or some other users) to easily see Walker Beach, looking south? What's wrong with the gallery wrapping responsively so I don't have to horizontally scroll? — JohnFromPinckney (talk / edits)23:51, 22 June 2021 (UTC)
That doesn't sound like it would work very well on mobile, or on narrow desktop screens/narrow browser windows. — Goszei (talk) 00:46, 23 June 2021 (UTC)
Looking further, I found the "perrow" parameter; it just doesn't seem to work for galleries in packed mode. @Goszei, for mobile/smaller devices, the galleries will definitely need to go onto multiple lines; the thing I'm trying to prevent is just the situation where 6 photos are on one line and 1 goes onto the next. {{u|Sdkb}}talk03:01, 23 June 2021 (UTC)
Uncaught ReferenceError: pathoschild is not defined
at <anonymous>:5:89
at domEval (/w/load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:11)
at runScript (/w/load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:13)
at enqueue (/w/load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:11)
at execute (/w/load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:14)
at doPropagation (/w/load.php?lang=en&modules=startup&only=scripts&raw=1&skin=vector:6)
when running this code:
pathoschild.TemplateScript.add({
name:'ScriptInstallation',
script: function(editor) {
editor
.prepend('== Installation ==')
.prepend(';Method 1:')
.prepend('Get [[User:Enterprisey/script-installer|ScriptInstaller]], then navigate to [[{{FULLPAGENAME}}.js]] and click "Install" at the top.')
.prepend(';Method 2:')
.prepend('(This can be used on any-language Wikipedia.)')
.prepend('# Place {{tlx|lusc|1{{=}}{{FULLPAGENAME}}.js}} on the bottom of [[Special:MyPage/common.js]] or [[Special:MyPage/skin.js]].')
.prepend('# Press "Publish Changes".')
.prepend(';Method 3:')
.prepend('(This can only be used on the English Wikipedia.)')
.prepend('# Place {{tlx|iusc|1{{=}}{{FULLPAGENAME}}.js}} on the bottom of [[Special:MyPage/common.js]] or [[Special:MyPage/skin.js]].')
.prepend('# Press "Publish Changes"')
.appendEditSummary('Added script installation text')
.clickDiff();
}
});
When I check my watchlist the Navajo line has an odd note beside it - a small clock symbol - and when I hover over it "30 days left on your watchlist" shows up. When did that start and why? Vsmith (talk) 16:14, 23 June 2021 (UTC)
Thanks, wasn't aware of that ... I'll avoid that twinkle bit or twiddle with prefs. I've got a huge watchlist that needs trimming ... tried once, but Wiki timed out trying to load it :) Vsmith (talk) 17:56, 23 June 2021 (UTC)
This is a great addition. There are loads of pages on my watchlist where I made an edit and wanted to see any follow-up and years later they are still there. Setting the watch to a week or a month is very useful. — Jts1882 | talk19:46, 23 June 2021 (UTC)
The mobile display of ambox, as well as certain other elements on the mobile version, is still basically being controlled by the Wikimedia Foundation's developers. I believe that your suggestion that we do here will be overridden with code that they control on the mobile version. MW:Reading/Web/Projects/Mobile Page Issues probably best describes what they have been specifically doing with ambox. MW:Reading/Mobile Friendly Content probably best describes what they are trying to accomplish overall: basically, the MediaWiki software was not initially designed with responsive web design by default, and many existing elements are only optimised for desktop browsers, so they still want to find ways make these things more mobile friendly. Thus, you might want to instead post your suggestion at MW:Talk:Reading/Web/Projects/Mobile Page Issues. Thanks. Zzyzx11 (talk) 02:01, 24 June 2021 (UTC)
Proposal: Show amboxes (article message boxes) by default on mobile devices
On desktop, amboxes are considered important enough to display even to non-editors. This is used by non-editors to recognize when there are potential issues with an article on WP. But on the mobile versions of Wikipedia, the amboxes are hidden or relegated to a subpage (E.g. on the Wikipedia iOS app, amboxes are not displayed but instead there is a link "page issues" at the end of the article.)
My opinion is, we should give amboxes similar prominence on mobile that we give them on desktop. I.e. display them at the top of the page or section. 15:44, 20 June 2021 (UTC) TOA The owner of all ☑️
The mobile display of ambox, as well as certain other elements on the mobile version, is still basically being controlled by the Wikimedia Foundation's developers. I believe that your suggestion that we do here will be overridden with code that they control on the mobile version. MW:Reading/Web/Projects/Mobile Page Issues probably best describes what they have been specifically doing with ambox. MW:Reading/Mobile Friendly Content probably best describes what they are trying to accomplish overall: basically, the MediaWiki software was not initially designed with responsive web design by default, and many existing elements are only optimised for desktop browsers, so they still want to find ways make these things more mobile friendly. Thus, you might want to instead post your suggestion at MW:Talk:Reading/Web/Projects/Mobile Page Issues. Zzyzx11 (talk) 02:01, 24 June 2021 (UTC)
Page tagged for speedy, did not get added to watchlist
I recently tagged a page for speedy deletion using those tabs at the top of the page (can't remember what they are called). The tagging happened correctly, the creator was messaged, but the page itself did not then appear on my watchlist. I'm sure when I've done this before the pages were added to my watchlist. DuncanHill (talk) 23:21, 22 June 2021 (UTC)
The question here is to address what we should general aim to try to do. I'm well aware each project is independent and can setup IRC channels wherever they want. However, we could decide that we encourage specific servers and discourage others, and try to migrate the 'official' Wikipedia/Wikimedia IRC channels to Libera instead of Freenode. Headbomb {t · c · p · b}20:17, 20 May 2021 (UTC)
I don't see that this needs an RFC? Wikimedia group contacts have already announced they will migrate to Libera on meta. Izno (talk) 20:22, 20 May 2021 (UTC)
This affects a lot more than what the WMF does. For example, there's #wikipedia-bag, #wikipedia-en-afc in templates like {{AfC welcome}} (including substed version of it), etc. etc. etc. Headbomb {t · c · p · b}20:41, 20 May 2021 (UTC)
So, why would we go somewhere else? Who has the knowledge in the community for that? Who wants to volunteer for that? Izno (talk) 20:46, 20 May 2021 (UTC)
See the above articles. As for who has the knowledge, there's busloads of technical users here that can help with this. Headbomb {t · c · p · b}20:48, 20 May 2021 (UTC)
Ok, let me spell it out then: A) We don't need an RFC. An RFC is a waste of the community's time on the point. B) Plain common sense is "go where WMF says they're taking the main channels". Izno (talk) 20:53, 20 May 2021 (UTC)
If I understand correctly, the channels were set up by the individual channel operators? So I suggest they can come up with a proposed plan and publicize it. I imagine most people will be fine with that, but in the event anyone objects, it can be discussed further. (Just as at meta, there may be interest in other chat tools, but that shouldn't stop any transition plan under these specific circumstances.) isaacl (talk) 21:06, 20 May 2021 (UTC)
Is there a setting to make the enhanced watchlist (that which groups changes to the same page) show up on mobile? Cheers. — 𝐆𝐮𝐚𝐫𝐚𝐩𝐢𝐫𝐚𝐧𝐠𝐚☎03:17, 24 June 2021 (UTC)
Special:WantedCategories and Special:UnusedCategories always update every 3 days, starting on the first day of the month but they haven't updated today. I have no idea what bot or program is responsible for updating Special pages and these pages have no edit history to check. They have talk pages but not ones that anyone ever checks. Any ideas? Is there a general lag today? LizRead!Talk!19:52, 25 June 2021 (UTC)
@Liz: this happens on the back-end by a cron job, the Job gets run every three days, but can take more than a day to get through the list. So unless it is a couple of days over, its not worth raising a phab ticket yet. — xaosfluxTalk20:10, 25 June 2021 (UTC)
@Liz: OK, just FYI this is now phab:T285538; which is marked 'resolved' - but please note that only means the underlying problem is resolved, not the instance of the problem. So the job that updates these is allegedly running now, but it may take time to realize the update. — xaosfluxTalk20:25, 25 June 2021 (UTC)
If anything is urgently needed, please poke someone and we can get a member of SRE to run it but otherwise it will just pick up on its next normal run. ~RhinosF1(Chat) / (Contribs)20:27, 25 June 2021 (UTC)
I thought once upon a time, we were able to just open the lead section and edit it. I don't see such a link now. Did I imagine such a thing existed? Did I change some preference that took it away? I use Modern skin. — Maile (talk) 02:22, 24 June 2021 (UTC)
Yes, it's there all right. And I apparently already had it checked. But in editing any article, that lead section link is not available. Not in Chrome, and not in Firefox. Hmmmm. — Maile (talk) 10:04, 24 June 2021 (UTC)
Works for me@Maile66: the gadget places the [edit] link at the end of the page title. Are you using the desktop site? Which skin are you using? — xaosfluxTalk10:19, 24 June 2021 (UTC)
I'm using Modern skin, and have been using it for years. What is the desktop site? — Maile (talk) 10:32, 24 June 2021 (UTC) Nevermind. It was there all along, but I was looking in the wrong place. — Maile (talk) 10:33, 24 June 2021 (UTC)
I can totally understand not seeing it. Putting the link next to the title is not an intuitive place for "click me to edit the lead". If I didn't know what it was, I would guess it was either "click me to edit the title" (i.e. rename the article), or "click me to edit the whole article" (i.e. a duplicate of the Edit link in the header bar). -- RoySmith(talk)14:32, 24 June 2021 (UTC)
@RoySmith: it does have a tool tip label of "Edit lead section", and all other sections have the edit link next to the header name. I suppose it could inject a new L2 header called something like "== Lead ==" and put it there maybe? If you have some good ideas for improvement, please feel free to follow up at MediaWiki talk:Gadget-edittop.js. — xaosfluxTalk14:42, 24 June 2021 (UTC)
insert a caption here
Ah, so it does. I never noticed that before. Perhaps instead of just the tooltip, it could alter the displayed text? -- RoySmith(talk)14:50, 24 June 2021 (UTC)
It currently makes use of a single message file there, likely so that it is translated automatically. Drop your suggestion at MediaWiki talk:Gadget-edittop.js and maybe someone will look in to it (or feel free to sandbox a new script if you would like). — xaosfluxTalk08:32, 26 June 2021 (UTC)
Yup, I commented there as well. Note, I'm not complaining. I mostly stick to the backend because it's a lot easier than UI :-) -- RoySmith(talk)16:05, 26 June 2021 (UTC)
Proposal for a bot to substitute uses of Template:Date in articles as a one-time run
I've filed a BRFA that proposes a task to substitute all uses of {{date}} in mainspace, given that (a) its documentation says that it should only be used within other templates, and (b) there is no need to use the template in articles, simply writing out the date serves the same purpose.
This is technically a cosmetic task, as the rendered wikitext wouldn't change - I would hope that some latitude can be extended given that it would be a one-time run, and is "enforcing" (for want of a better word) existing documentation. Of course, we may decide that the documentation doesn't accurately reflect consensus - in that case the BRFA can be withdrawn/denied and the documentation updated.
It may be best for comments to be made at the BRFA to keep things in one place, although I'm equally happy for the discussion to happen here. Thanks! firefly ( t · c ) 17:20, 27 June 2021 (UTC)
Not fundamentally possible without resorting to CSS in the task center page itself. I do not think you want to do that. Izno (talk) 21:55, 27 June 2021 (UTC)
Is there a reason I wouldn't want to do that? This problem of trying to add elements to the side without disturbing elements in the center keeps on coming up, and I haven't memorized exactly how to solve it but I'm pretty sure it's solvable. {{u|Sdkb}}talk22:52, 27 June 2021 (UTC)
If you just want to move shortcuts before the header, move the div with it before the header, and put {{clear}} (or apply the style clear to the div) just after the div.
If you want to put shortcuts to float on top of the header, use position:absolute (with z-index:1 to make it on top). MarMi wiki (talk) 23:43, 27 June 2021 (UTC)
Crosswiki notifications and the responsive monobook skin
I use the responsive monobook skin, which changes several things depending on the size of my window/screen. When I receive a crosswiki notification, it shows up as a little blue "1" in the notifications inbox in both designs. However, if I click on it to see the notification, I can only see it in the classic design. There is no "notifications from another wiki" message in the small-screen/mobile version (but all notifications from enwiki are visible there). There seems to be no way to find out what a crosswiki blue 1 is about without changing screen size / turning off the responsive design. Is this intentional? If no, can this be fixed? If yes, why? Could crosswiki notifications at least have a different colour? (So I know I can't read them on my phone and will have to wait until I get back to my desktop). The software telling me I have messages but sometimes not telling me what they are isn't great. —Kusma (talk) 08:57, 19 June 2021 (UTC)
The problem also occurs in Vector, actually. The only information about crosswiki notifications is in the "Recent activity" box, which disappears without a trace when the screen/window size is too small. So if you get a crosswiki ping, you can't tell where it came from and you can't clear it from your inbox. I dpn't quite know where to report this. —Kusma (talk) 18:52, 24 June 2021 (UTC)
Adding .mw-echo-ui-notificationsInboxWidget-sidebar {display:block;} to User:Kusma/monobook.css unhides the Recent activity box. It is a rather ugly workaround (I have to scroll to the side on my phone), but I don't speak CSS so I can't fix this. Thank you Rummskartoffel for the help! —Kusma (talk) 14:51, 28 June 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Wikisources have a new OCR tool. If you don't want to see the "extract text" button on Wikisource you can add .ext-wikisource-ExtractTextWidget { display: none; } to your common.css page. [31]
Problems
You will be able to read but not edit the Wikimedia wikis for a few minutes on 29 June. This is planned at 14:00 UTC. [32][33]
Changes later this week
The new version of MediaWiki will be on test wikis and MediaWiki.org from 29 June. It will be on non-Wikipedia wikis and some Wikipedias from 30 June. It will be on all wikis from 1 July (calendar).
Future changes
Threshold for stub link formatting, thumbnail size and auto-number headings can be set in preferences. They are expensive to maintain and few editors use them. The developers are planning to remove them. Removing them will make pages load faster. You can read more and give feedback.
A toolbar will be added to the Reply tool's wikitext source mode. This will make it easier to link to pages and to ping other users. [34][35]
I don't know if it will help, but I want to draw VPT editors' attention to the first item under "Future changes". The proposal is to eliminate custom the thumbnail size preference for editors (under Prefs - Appearance). The phab task states that only 4% of editors active in the last 30 days have this preference set, and it makes pages load slowly for us, so they want to get rid of it. See T284920 for details. – Jonesey95 (talk) 20:47, 28 June 2021 (UTC)
The Wikipedia logo shown on the left-hand-side panel with Skin:Timeless is a PNG image, but all other Wikimedia projects with Skin:Timeless use an SVG image for their logo. As the PNG logo is bitmap, the Wikipedia logo becomes blurry on higher-resolution displays, yet the SVG logo of Wikipedia is available. I also noticed that this issue happens across all Wikipedias so there should be a global solution for it. 🐱💬11:32, 28 June 2021 (UTC)
At the current FARC for Duke University, I'm wondering if there's any easy way to simultaneously ping all 50 members of Category:Wikipedians by alma mater: Duke University. I've seen this done fairly often at Wikidata, but I'm not sure if the same functionality exists here. If not, I think it'd be a nice thing to create. We would probably want to build a safeguard to prevent abuse by limiting it to categories with, say, less than 100 members. {{u|Sdkb}}talk20:18, 28 June 2021 (UTC)
What is easily doable is something like the following:
letapi=newmw.Api();awaitapi.get({action:"query",format:"json",list:"categorymembers",cmtitle:"Category:Wikipedians by alma mater: Duke University",cmnamespace:"2",cmlimit:"50",formatversion:"2",}).then(response=>response.query.categorymembers.map(entry=>`[[${entry.title.split("/")[0]}]]`).join(", "));
That leaves you with a string like this: [[User:Example]], [[User:Example]], [[User:Example]] ..., which you can then just paste as wikitext. – Rummskartoffel21:21, 28 June 2021 (UTC)
I usually just paste these kinds of things into my browser console ad-hoc, but I've thrown together a quick user script that adds a link to the portlet (left sidebar, at least on Vector and Monobook) under "Tools". If you click it when on a category that contains users, it'll pop up an alert box with the generated wikitext, from where you can copy it. It's not very pretty, and if you click it elsewhere, it'll probably break, but it should get the basic job done unless your browser's from the stone age. – Rummskartoffel22:37, 28 June 2021 (UTC)
@Rummskartoffel, thanks so much for putting that together! It doesn't seem to make anything show up for me, but if there's not an obvious bug, don't worry about it; the chance of reaching anyone through this willing to save Duke from delisting is pretty close to zero no matter what. {{u|Sdkb}}talk03:29, 29 June 2021 (UTC)
@Sdkb: That's because you didn't install it correctly 😜. Your common.js tries to import the documentation page (thanks for creating that, by the way), not the script itself. Try importScript("User:Rummskartoffel/generate pings.js"); instead. – Rummskartoffel08:48, 29 June 2021 (UTC)
Oops, yeah, that's the one I meant; thanks for catching and finding the correct link. There's a lot of problems on phabricator these days haha {{u|Sdkb}}talk23:45, 28 June 2021 (UTC)
With regards to upper limits: there's a maximum number of people to ping built into the notification system. It just barely works for this category of 50, but any attempt to ping 51 people would fail (per mw:Manual:Echo) Vahurzpu (talk) 23:08, 28 June 2021 (UTC)
How would an invitation to participate in a WP:FAR discussion produce "fragmented discussion"? Just my 2c: pinging doesn't leave much of a paper trail and seems like a sneaky way to canvass; if you must canvass, might as well be open about it. -FASTILY23:00, 28 June 2021 (UTC)
On the contrary, pinging is visible exactly where the discussion is taking place. Leaving 50 messages on talk pages may make people reply to the message on their talk pages instead of at the centralised discussion. —Kusma (talk) 23:05, 28 June 2021 (UTC)
But for "paper trail" reasons, the OP's suggestion to be able to ping the members of a category should probably not be implemented, unless there is a way to find out category membership at a given time in the past. —Kusma (talk) 23:11, 28 June 2021 (UTC)
Incorrect timeframe in notice
Where is the message that currently displays "Technical maintenance will be performed soon 05:00 UTC - 05:30 UTC During this time you might not be able to save any edits."? It's wrong; 05:00 UTC was more than eight hours ago. Judging by Tech News above, it should read 14:00 UTC to 14:30 UTC - can we get this changed? --Redrose64 🌹 (talk) 13:40, 29 June 2021 (UTC)
Fixed, sorry for inconvenience, and thanks for noticing. Do bear in mind that in practice, the read-only may last up to... 2 minutes, just as it was last year :) SGrabarczuk (WMF) (talk) 13:51, 29 June 2021 (UTC)
Why browsing wikipedia is much, much slower when logged in?
I have noticed this for quite a while so decided to do a quick test with it today (it's not meant to be super comprehensive so bear with me).
I open Phillip Davey with my dev tool opened and "Disable cache" checked. I then refresh the page 10 times to see the average loading time. Then I repeat it after logged out.
The difference is very obvious: ignoring any resources, only look at the first HTTP request for the HTML:
Logged in: the average time is 520 ms.
Logged out: the average time is 15 ms.
I understand it can't be as fast, but this is pretty bad and very noticeable when just browsing around. Is there anyway to improve it? Thanks! --fireattack (talk) 14:47, 28 June 2021 (UTC)
Why would you keep the "Disable cache" checked? When you're logged in, more resources (such as gadgets you've enabled) are loaded so uncached loading time is expected to be larger. But the difference will be that much only on the first load if caching is enabled. Also, there are a couple of preferences (see phab:T284917, phab:T284920, phab:T284921) that disable server-side caching and greatly slow down page loads. – SD0001 (talk) 15:01, 28 June 2021 (UTC)
Except... I can disable the global setting of thumb sizes, but can't disable it at enwiki; can only choose one. What's this referring to?
An ad-hoc analysis among users of en.wiki.x.io who have edited in the last 30 days shows that about 4% of active editors have the "thumb size" option set.
@SD0001: to make the test more consistent. The result isn't much different without it, though, since the HTTP request for the page itself won't be (browser-side) cached anyway (it mainly affects the resource like JS or media). I checked the preferences you linked, I didn't enable any of them (except for the thumb size, which I didn't find a disable option. I use the default 220px). --fireattack (talk) 03:36, 29 June 2021 (UTC)
Caching in general is not available for logged-in users. Some preferences (in addition to SD0001's) and all the gadgets you can enable, as well as simply not being set up for it from a data center level (that's mediawikiwiki:Wikimedia Performance Team/Active-active MediaWiki, which means your traffic goes to the singular data center equipped for logged in edits). Izno (talk) 16:17, 28 June 2021 (UTC)
Yes, for many operations your request has to travel all the way to the USA, which is a much longer/slower path for those far outside the USA than just getting some cached data from a geo localised caching center. —TheDJ (talk • contribs) 10:27, 29 June 2021 (UTC)
This is a long-standing known issue due to the traditionally monolithic mediawiki architecture- all operations, including reads that are not cached, have to go to the primary datacenter. There is an Epic task T270223 that will mitigate that, talking the first step towards being able to serve read-only traffic from a remote (closer to you) datacenter. This, however is a complex issue, that has been prepared for a long time and will require a lot of work to be made possible (for example, what looks as a read only request sometimes gets converted to read-write, or has to write its cache, and consistency has to be kept between multiple geographically distributed locations)- as it requires new hardware, changing the application request workflow, and the data architecture for the application to be more "distributed".
However, being able to make the site faster for contributors is the main goal of such a project. One thing that may not be seen is that anonymous browsing (readers without account) has been made much faster in the last years as more CDN trafic caching sites have been opened- and now it should be the turn of improving the experience for logged-in users, too. The Performance team is hard working at making this possible. Some extensions may need adjustments to take advantage of this and you could help with feedback and code! --JCrespo (WMF) (talk) 15:10, 29 June 2021 (UTC)
Wikipedia displaying as text only when Adblock enabled
@Satellizer: Not only that, there's absolutely no point in having ABP enabled for any wikimedia site - not long after installing it, I switched it off for en.wiki.x.io, en.wiktionary.org, commons.wikimedia.org, meta.wikimedia.org and several others because they never run adverts in the accepted sense - our occasional fundraisers are easily dismissed, and things like yesterday's notice about the read-only period I do need to see. Every time I see that red octagon on a Wikimedia site (such as those), it gets clicked and disabled. --Redrose64 🌹 (talk) 07:27, 30 June 2021 (UTC)
At WikiProject Sweep, we're still stuck at the stage of trying to determine the criteria of the list of pages to be swept. Would anyone be able to answer or know where we could go to find the results of these queries?
What percentage of articles created prior to 20 September 2012 have been edited by at least 3 extended-confirmed editors who have made a non-minor edit? What does a sample of these pages look like?
What about for 5 EC editors, 10 EC editors, or 25 EC editors?
@Sdkb I'm not sure I understand the criteria. I assume you mean, "editors who were EC at the time they made their edit". If so, that sounds like a very expensive query, since you'd have to re-evaluate each editor's EC-ness as of the time of each edit. "Once EC, always EC", so you could do some caching, but, my first guess is this is probably still an intractable problem.
It would be hugely more efficient if the criteria were "editors who are EC now". I assume you're using EC as a proxy for "is trustworthy". Let's assume that somebody who's trustworthy now was always trustworthy (even if they hadn't accumulated the editing history to demonstrate it yet). According to WP:EDITORS#Number of editors, there's only about 57k EC users (not sure if that includes admins; if not, then add another 1k), so if we can make that assumption, it would be a huge optimization.
Reality check: there's 41M accounts, and 57k EC users. So only about 0.1% of accounts ever reach EC? Is that plausible? -- RoySmith(talk)15:35, 29 June 2021 (UTC)
@RoySmith, you're correct; I'm using EC as a proxy for "experienced enough to know to tag/AfD the article if it's clearly warranted." EC at the time of the edit would be ideal, but if that's not possible, EC overall would seem like a decent alternative—except for the issue that @Rummskartoffel pointed out. We don't want to exclude active editors who retired before 2016, which could be quite a few of the main contributors to WP's earliest articles. Maybe we could trying using autoconfirmed instead raise the numbers to 5, 10, 25, and 50? Sorry I'm not able to give a single precise request for the query; the thing about this one is that it's hard to tell exactly what we want until we've seen the result with a few different settings and examined whether it seems too wide or narrow. {{u|Sdkb}}talk06:53, 30 June 2021 (UTC)
You may want to exclude some of the top script/AWB users from your queries. I would expect that thousands of articles have been touched by Ser Amantio, Koavf, Magioladitis, BHG and Rich Farmbrough, making edits that are "minor" in some sense but not necessarily minor in the sense of the checkbox. I don't expect that any of them vouches for the notability of all articles they touch. And that's fine. It just means that there is no simple and reliable automatic way to say "this article has been looked at" just by ECness of the editors touching it. —Kusma (talk) 08:58, 30 June 2021 (UTC)
Interesting, I didn't know that some of those aren't marked as minor. I believe all semi-automated edits are marked with a tag, and hopefully that has always been the case, in which case excluding those edits should be possible. {{u|Sdkb}}talk13:32, 30 June 2021 (UTC)
For what it's worth, the help page states we shouldn't mark category changes as minor, so if you follow that HotCat shouldn't (standard caveat that it's not a policy or guideline etc). There'll also be edits that should've been marked as minor but couldn't, such as those made from the mobile website. Sdrqaz (talk) 13:39, 30 June 2021 (UTC)
Yeah, I would exclude anything marked as minor or bot (do bots earn autoconfirmed???), plus anything we can recognize as somebody on a gnoming mission like updating cats, adding short descriptions, etc. Most of these things are done in a semi-automated way with some tool support, and the tools usually add a recognizable comment to the edit description (like HotCat does). -- RoySmith(talk)13:47, 30 June 2021 (UTC)
Would it be possible to perform some sort of check so that {{Contains special characters}} is only displayed when someone actually lacks the rendering support needed to display them? It's unneeded otherwise, and as rendering technology has improved over time, this has become increasingly common. {{u|Sdkb}}talk06:34, 30 June 2021 (UTC)
There's an outside chance this is related to T270209. If you know how to examine your browser console, could you open it up and see if you get the same kind of errors as noted in that ticket? -- RoySmith(talk)15:05, 30 June 2021 (UTC)
With so many examples being available on commons - I'd open a phav ticket. Just put it in simple language, but provide clear examples - someone will clean up the tech details. — xaosfluxTalk15:15, 30 June 2021 (UTC)
Cross origin errors when accessing Wikidata
I'm trying to query Wikidata in an enwiki gadget. Sadly it just throws CORS errors, the "Access-Control-Allow-Origin" header is missing. I don't get it.. does this mean it's impossible to query Wikidata in an enwiki gadget? — Alexis Jazz (talk or ping me) 21:16, 27 June 2021 (UTC)
Nardog, I think I finally have it working. I added origin=https%3A%2F%2Fen.wiki.x.io to the URL. mw.ForeignApi should do something similar but screws up the script in other ways. (if I remove the await it breaks, if I don't the script won't load) So I'll just stick with origin=https%3A%2F%2Fen.wiki.x.io. Don't understand why this has to be specified. Like evilmiddleman.com can't add origin=https%3A%2F%2Fen.wiki.x.io to the request URL because that would be lying which would be an insurmountable ethical hurdle for evilmiddleman.com? Well whatever, it works now. — Alexis Jazz (talk or ping me) 22:55, 27 June 2021 (UTC)
@Alexis Jazz evilmiddleman.com would need to add origin=evilmiddleman.com, not origin=https%3A%2F%2Fen.wiki.x.io. In which case, the API will refuse the request since evilmiddleman.con is not on the CORS whitelist. On WMF wikis (which are on the CORS whitelist), you can just add origin=*. There's a phab ticket somewhere that seeks to make this requirement unnecessary. – SD0001 (talk) 08:14, 28 June 2021 (UTC)
This is documented here: Manual:CORS It's basically there because otherwise we would have to Vary on each and every api request of the Origin header, and especially because we need to vary on authenticated and non-authenticated responses of cross site requests (or something like that). —TheDJ (talk • contribs) 08:19, 28 June 2021 (UTC)
TheDJ, I don't quite understand. If the origin parameter is compared to something else (and it appears it is as visiting an api request with defined origin directly results in the error 'origin' parameter does not match Origin header), why doesn't it just check that something else? A passed parameter can't be trusted anyhow. — Alexis Jazz (talk or ping me) 19:12, 30 June 2021 (UTC)
Grouping references
Hey there,
I've a contentious statement that is supported by a lot of sources, and I'm trying to group them in a note so we don't have a whole list in-line. Is there a better way to do this than this? Specifically, is there a way of exposing the refs in the note, so that they're immediately visible when hovering with the mouse pointer?
HighInBC, it looks to me as though in the case linked the template simply wasn't filled in properly; the demonstration value of blocking administrator was left in place. I've tried using the template and filling in the parameters and it appears to work as expected. I could however be missing something! firefly ( t · c ) 06:25, 1 July 2021 (UTC)
HighInBC, in the template code the accept template uses {{{3|{{{reason|original unblock reason}}}}}} (line 23) to get the unblock reason, which is correct, but the decline template uses {{{1|{{{reason|original unblock reason}}}}}} (line 27) to get the unblock reason, which is wrong as parameter 1 is where the blocking administor's name goes. To fix this, you would just need to replace the shown code on line 27 with the shown code on line 23. – BrandonXLF (talk)06:44, 1 July 2021 (UTC)
Thank you Firefly and BrandonXLF for the help. I have tested it in my sandbox and it seems to have worked. I implemented the change in the temple itself[36]. Oh lawd I hope I did not break the entire unblock system! HighInBCNeed help? Just ask.07:01, 1 July 2021 (UTC)
I keep getting an error message, "An error has occurred while searching: Search is currently too busy. Please try again later." In other tries it is super slow. Abductive (reasoning)02:33, 30 June 2021 (UTC)
A couple of days ago, much of the server infrastructure behind wikipedia was moved from one data center to another, which might account for transient performance problems like this. If these failures still happening for you, a phab ticket should be opened (if you don't have phab access, post here and I can open the ticket for you). -- RoySmith(talk)11:45, 1 July 2021 (UTC)
Category:Articles containing undetermined-language text
Hey! So I was sending a mass message, and while my log states that I only sent one message, why does MediaWiki MassMessage Delivery sometimes send the message later again, to certain pages on the message list, but not to all pages, for example, I sent a message at X o clock, only once. Then a few hours later, I did not send another message, via mass message, and the same message will get sent again. Even though I only sent it once, and my log only shows I sent it once. This has happened 2 times now. Is there a way to fix this? --つがるTalk to つがる:)🍁21:41, 2 July 2021 (UTC)
@つがる: can you be more specific in your examples? I see you sent a message here: LogId:119014629. Can you point to the duplicate log entry, or the edit made by the service you don't think you have occurred? — xaosfluxTalk22:26, 2 July 2021 (UTC)
The OpenStreetMap conference, "State of the Map", has an impressive, easy to use, tool for switching timezones on its programme page.
Could we use that, or something very like it, on Wikipedia and sister projects, for our events? The tools I've seen used by us for such things seem much more clunky. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits10:52, 3 July 2021 (UTC)
Hi, If the articles have been deleted uner PROD or BLPROD, where can we find the hist diff/log for the nomination of PROD/BLPROD edits? Thaks in advance. Cassiopeia(talk)09:32, 3 July 2021 (UTC)
Redrose64 Thank you for your quick reply. May be I rephrase my question. I wan to find the hist diff of the nomination edit of a deleted PROD/BLPPROD page. since the article is deleted, which means the history page of the deleted page is also deleted. Could we able to find the hist diff anywhere else besides the creator talk page where we place a notification of the PROD/BLPPROD notification? Thanks in advance. Cassiopeia(talk)10:11, 3 July 2021 (UTC)
@CASSIOPEIA: Non-admins can see metadata of deleted revisions (via API or Quarry) – this is exposed via deleted-metadata-link.js. Since PRODs are usually done via twinkle, you can usually guess the edit in which the PRODing occurred by looking at edit tags. If the PRODing occurred via PageTriage, they're also recorded in the page curation log. – SD0001 (talk) 16:22, 3 July 2021 (UTC)
SD0001 Thank you for the info above. The thing is I am not a technical editor and not sure how to use and where to find the API/Quarry to find the PORD info by looking at the edit tag. I would appreciate if you could provide step by step instructions by using Bent Creek Country Club deleted page PRODing via Twinkle. Thanks in advance. Cassiopeia(talk)02:21, 4 July 2021 (UTC)
@Cassiopeia: Just click on that red link. Above the red box, you should see the text: "See deleted revisions: last edited by Nearlyevil665 at 17:48, 20 June 2021 (UTC) Look up snippet in PROD grid". Click on "deleted revisions" to see further info. The latest edit with tags: [ "twinkle" ] is probably the edit that added the PROD. If you also want the diff ID for some reason, you can look for &drvprop= in the URL and replace it by &drvprop=ids%7C. – SD0001 (talk) 05:07, 4 July 2021 (UTC)
SD0001 Thank you for your quick reply. I assume the "red link" you mentioned above was Bent Creek Country Club. However, what I get is this and no "See deleted revisions: last edited by Nearlyevil665 at 17:48, 20 June 2021 (UTC)" or anything your mentioned above. Pls advise. Thanks in advance. Cassiopeia(talk)05:23, 4 July 2021 (UTC)
@Cassiopeia That's weird, since you do have User:SD0001/deleted-metadata-link.js in your common.js. Check your browser console (see WP:JSERROR #6 on how to open it) to see if there are any errors in red. – SD0001 (talk) 05:32, 4 July 2021 (UTC)
This may be related to this discussion at the RedWarn talkpage.
When I try to edit pages, I cannot publish changes, which I think is due to this error:
mdlLogic.js:1 Uncaught URIError: URI malformed
at decodeURIComponent (<anonymous>)
at Function.t.EPPyTH (mdlLogic.js:1)
at t (mdlLogic.js:1)
at Object.mw.loader.load (<anonymous>:2:180)
at Object.preloadDeflate (<anonymous>:351:854)
at <anonymous>:40:348
As I said in § Scripts only working on certain pages, this error occurs in RedWarn code, so I recommend you ask the RedWarn developers. If you want to be sure whether it's really RedWarn that's stopping you from publishing, temporarily uninstall it and see if the problem still occurs. – Rummskartoffel19:40, 3 July 2021 (UTC)
I am with you one of the Arabic Wikipedia editors :), I work sometimes in a dark atmosphere (like now when writing to you) on a bright white screen, which annoys me a bit especially when I see the page of the site, I think we could coordinate with the rest of the other wikis or with programmers in Wikipedia in order to give this appearance, and make it Within the Appearance section of the preferences page for each Wikipedia user, with the ability to specify the time of their appearance (as it is present and familiar to you in the settings of computers (such as Windows 10) or mobile phones), and it is not hidden from you that these sites exist in other recent sites such as Twitter and YouTube.
The problem is that you have two sources with the same author and year (the first bulleted entry in the sources list and ref 15), so the sfn doesn't know what to link to. The easiest solution is probably to add |ref=none to the wikitext of ref 15, to explicitly state that the sfn should link to the other one. * Pppery *it has begun...21:06, 3 July 2021 (UTC)
@Jo-Jo Eumerus: As noted by Peppery, there are two similar sources:
<ref name="CantwellNewman2016b">{{cite report|last1=Cantwell|first1=Kasey|last2=Newman|first2=Jim|url=https://repository.library.noaa.gov/view/noaa/17556|year=2016|title=Okeanos Explorer ROV dive summary, EX1605L3, June 26, 2016}}</ref>
* {{cite report|last1=Cantwell|first1=Kasey|last2=Newman|first2=Jim|url=https://repository.library.noaa.gov/view/noaa/17555|year=2016|title=Okeanos Explorer ROV dive summary, EX1605L3, June 25, 2016}}
They're not identical: the one used as a ref has a URL ending in 17556 and a date of June 26 whereas the one in the Sources section has 17555 and June 25 respectively. The three instances of {{sfn|Cantwell|Newman|2016|p=3}} don't know which to link to.
If they relate to the 17556/June 26 one, remove the one in Sources.
If they relate to the 17555/June 25 one, add |ref=none to the one in References.
Alternatively, you could alter |year=2016 to |year=2016a in one and to |year=2016b in the other, and in the {{sfn}} alter |2016 to either |2016a or to |2016b whichever is applicable. --Redrose64 🌹 (talk) 22:33, 3 July 2021 (UTC)
OK, did this with ref=none. I didn't realize that sfn wouldn't be able to see the letter at the end of the sfn template. Jo-Jo Eumerus (talk) 07:58, 4 July 2021 (UTC)
Admins can't delete large-revision pages?
If [http://en.wiki.x.io/wiki/Talk:Main_Page?action=delete] this is correct, I have a question. Does that mean that if I were to make over 5,000 edits to my user page, no one here would be able to delete it? And what if over 5,000 edits were made to an admin's own user page, and they couldn't delete it by their own request? This doesn't seem clear why the protection should be given in this case. Are there any exceptions to this? And can an admin send a screenshot of what happens when attempting to delete pages with 5k+ revisions? 54nd60x (talk) 12:26, 3 July 2021 (UTC)
@Jo-Jo Eumerus: This doesn't seem right. That means that approximately 5k+ articles cannot be deleted by anyone here. I also noticed that the $1 in the message seems to vary across different wikis, so why did we decide to set the limit there so that it would impact so many articles? 54nd60x (talk) 06:12, 4 July 2021 (UTC)
No, the limit is 5k on every WMF wiki. The number might be put into a different perspecive in the translated messages (english is the original), but the limit itself is the same. It is an performance decision (as in performance of the server) and not a big deal either, the stewards will still make the same action as an admin would.--Snævar (talk) 06:53, 4 July 2021 (UTC)
Apparently the 5k limit is not all. There is also an special deletion in place when pages exceed 1000 revisions. When an page that big gets deleted, that is put into Job queue and the revisions are deleted in batches, from oldest to newest. Even if the page gets edits or even moved from the point that the admin requests for deletion, then that action is still archived just like with an normal deletion, and any subsequent deletions of the same page whilst this process is underway does not affect the process - that was actually tested by the developers. There is a note on pages like those that there is an deletion in progress. See phab:T198176.--Snævar (talk) 10:38, 4 July 2021 (UTC)
If my memory serves me right very large pages were deleted in the past, either by accident or due to compromised accounts. It was highly disruptive causing a lot of load on the database servers as well as taking a long time to restore. This was very long ago, back when being here for 2 years made you an old timer. HighInBCNeed help? Just ask.08:52, 4 July 2021 (UTC)
I have problems with mobile search on Chrome in iOS, with the articles I frequently read not appearing with incomplete search strings, but appearing only when complete. For example, when typing in COV and expecting to see COVID-19, COVID-19 pandemic and related articles in results, I would see Covariance, Coventry, Covina, etc. Is this a bug, problem with browser, or something else?-TagaSanPedroAko (talk) 07:07, 3 July 2021 (UTC)
More information is needed I think. Have you searched for "Cov" before and then clicked on an covid search result ? The mobile search will remember past searches and make decisions on cases like that (it is a fairly recently added feature). I do not think it takes popularity in consideration (that is, ordering results by pageviews), which would work in this case.--Snævar (talk) 08:52, 3 July 2021 (UTC)
Yes. When I type the first three letters of COVID, I would expect COVID-19 and related entries should show up, but now, they only appear if I enter the whole thing.-TagaSanPedroAko (talk) 09:04, 3 July 2021 (UTC)
Ok, so the search does in fact order by traffic of pages and the history boosted search is an Android app thing. It should still show COVID-19 ahead of covariance tho. Filed a bug at phab:T286111.--Snævar (talk) 13:21, 3 July 2021 (UTC)
@Snævar: It’s not only COVID-related articles are affected by the bug. It also affects every article I frequently read both when logged in and logged out, that when I partially enter the name (or redirect) of an article I frequently read, I can’t see it until I type the whole term. I’ll give further examples. When I first enter the word “heart dis” and expecting to see Cardiovascular disease, I no longer see it unless I type the whole search term “heart disease”. Same also with “ill”, where I’m expecting “illegal drug trade” to appear in third or so place in results, but I see other articles with the first letters “ill”, the Ill disambiguation page, and the illegal drug trade in Colombia pages instead. It appears the issue is new; I haven’t encountered that with previous partial search entries in the last few days.-TagaSanPedroAko (talk) 15:09, 3 July 2021 (UTC)
I am also having the same issue with search bar. When I type something it no longer shows up (believe it's called autocomplete) in the upper right hand search bar/box. Is someone at Wikipedia working on fixing this problem? 158.222.185.250 (talk) 22:16, 3 July 2021 (UTC)
Search text problem
The past few hours if i type a name into the upper-right search bar it doesn't show up. For example if i type Julian Assange his name doesn't appear but "Julian Assange Show" does. are other users experiencing this weird issue?? 158.222.185.250 (talk) 11:55, 3 July 2021 (UTC)
It works for me. Did you have a space after "assange"? If so, then the title which matches the space (the show) will appear first but "Julian assange" [sic] should be the second option. Certes (talk) 12:14, 3 July 2021 (UTC)
I browse using Edge not Chrome. I'm still experiencing the problem, is Wikipedia working on fixing this autocomplete text issue with the search bar? 158.222.185.250 (talk) 22:10, 3 July 2021 (UTC)
The issue affects other browsers as well, I was able to reproduce it on Firefox. There is no upper right search bar on mobile web (it is in the upper center), so this is an desktop search issue here.--Snævar (talk) 07:22, 4 July 2021 (UTC)
Problems with autocomplete in search
I also want to reiterate problems I've been having with the search function today, as users here and here have. One way the problem manifest itself to me recently was that it required not only a complete string to return the article I was looking for; it required case sensitivity as well. For example, if searching for the article Light of a Clear Blue Morning, typing "light of a clear blue morning" or even "Light of A Clear Blue Morning" does not return it in the search bar. It must be complete and correct title case for the article to appear at all. This is not the only problem I have encountered with search today, but it is a specific anecdote that might shed some light on the nature of the problem.
It does not appear to be limited to Chrome on iOS, as has been previously speculated. This problem occurs for me across two devices, neither of which match those specifications.--Sunshineisles2 (talk) 23:22, 3 July 2021 (UTC)
There were some changes announced to search in m:Tech/News/2021/23. I don't think it would account for the behavior described here, but... The change described in T219550 does mention that it changes how case is handled. -- RoySmith(talk)23:51, 3 July 2021 (UTC)
I am also here to complain about an apparent negative change in how search and case interacts. When there isn't a redirect already created for a difference in case from the target title, the article no longer seems to appear, as OP describes. — Goszei (talk) 00:26, 4 July 2021 (UTC)
On Windows Chrome, by the way. I've also noticed other autocomplete abnormalities that I don't think were there before. Example: when searching the string "Barack Obama", the article does not appear in results until the last letter is entered (a total failure of autocomplete?) — Goszei (talk) 00:38, 4 July 2021 (UTC)
Yes, encountering this issue since the past few days. The search wants the complete string, you don't get suggestions on an incomplete string or case changed ones. Making the search suggestion useless. Changing to "Classic prefix search" in prefs does fix this though. Gotitbro (talk) 00:46, 4 July 2021 (UTC)
Remdesivir
Every version of Remdesivir, since creation, when I start to edit it, displays:
Lua error in Module:Sanctions at line 60: attempt to index field '_topicData' (a boolean value).
Done using AWB to replace, remove or {{tl}} out uses as appropriate, and so I've re-removed the covid topic area per GN above. Will double check for any Lua errors with this text, but I should not have missed any non-subst'ed uses. Dreamy Jazztalk to me | my contributions21:25, 4 July 2021 (UTC)
Kusma, (edit conflict) hi. I had used the search function to search for this error when I removed the covid gs from the data template. Obviously it took some time for pages to catch up and my search had not found all uses. The edit to remove the covid gs from the data template was because it has been superseded by WP:COVIDDS. Uses of the the GS editnotice with the covid topic need to be replaced with {{COVID19 DS editnotice}} or {{ds/editnotice}}. I will do another search for this error and fix any occurrences I find. In the case of this article the expiry parameter was set to a month in an edit in April last year, however, I think this editnotice should stay so I'll replace it. Dreamy Jazztalk to me | my contributions20:20, 4 July 2021 (UTC)
Sounds good. Part of the reason I pinged you was that I have no clue what the state of play is with Covid sanctions and so was hoping you'd know better what to do :) —Kusma (talk) 20:35, 4 July 2021 (UTC)
Bug related to the timeline functionality
In the article President of Portugal I saw some texts overlapped together at the Graphical timeline (since 1910) section, which illistrated with timeline tag, is there any way to make collision avoidance there? --Great Brightstar (talk) 17:40, 4 July 2021 (UTC)
Context: [38]. Google Chrome has implemented this "text fragment" URL feature for a year now, and it is today used within at least 9,260 articles. When a URL with this ":~:text=" string is clicked in Chrome, the page will be jumped to the corresponding text (strangely, my search shows these also have been added in inter-wiki links). They don't work in browsers other than Chrome, and don't even work all the time in Chrome (at least for me).
Google Search has been appending this string to its results in order to jump people to the text in webpages visible in the results snippet. It appears that editors have copy and pasting these URL's and leaving the extraneous string in; a small sampling of my search shows that they almost always have no correspondence with the text that is being verified, and so I think editors have been mostly been leaving this in the URL by complete accident. For this reason, I think they should be removed en masse, perhaps a task handled continuously by a bot (or AWB?). I would be given some pause if they were being used like a citation template's |quote parameter, but that doesn't appear to be the case at all, with almost all uses being apparently accidental.
Probably not much. As far as I can tell, these text fragment links degrade harmlessly to normal links to the intended webpage if the browser does not handle them, or the target webpage has since changed. It would be worth adding a note to our guidelines about internal links to make it clear that text fragment links are vulnerable to changes in the text of an article and internal links should always use ID references as the target. It might also be worth adding a note of some sort in other guidelines about references and external links — GhostInTheMachinetalk to me09:48, 3 July 2021 (UTC)
In theory, this is a wonderful feature that could be very helpful to our readers. Firefox has an add-on for it, and I hope that other browsers will implement it soon. If links are being created to the wrong point on the page, those individual links need attention, but let's not throw the baby out with the bathwater. The worst that can happen is to position the window at the wrong point in the page, which compares with the legacy behaviour of always showing the top. (I usually rant against browser-specific features, especially from the dominant supplier, but this one seems portable, beneficial and harmless.) Certes (talk) 12:22, 3 July 2021 (UTC)
Seems to be an optional (non-normative?) option for URLs, see [39]. Given that its use appears to be on the increase, is it worth putting a check in {{cite web}} that detects if the URL has a text fragment and asks the editor if they really need it?--Verbarson (talk) 20:27, 3 July 2021 (UTC)
Sorry, sloppy thinking. The check needs to be in the Source Editor function that prompts for and constructs the {{cite web}} template, not in the template itself.--Verbarson (talk) 08:13, 5 July 2021 (UTC)
Disappeared languages
I noticed that on the desktop version of vec.wiki are disappeared the languages links to the other Wikipedia languages editions. How can I solve? --62.18.11.223 (talk) 20:08, 2 July 2021 (UTC)
The language links still exist, but they have been moved; you can find them on the right, near the top of the page, just opposite the article title. Here is an image with the position of the menu visually highlighted. This change is a feature of the updates to the Vector skin being done as part of the Desktop Improvements, which Venetian Wikipedia is an early adopter of. If you have an account, you can switch back to the old version in your preferences. – Rummskartoffel21:01, 2 July 2021 (UTC)
(edit conflict) it looks like vecwiki is using a "new" version of Vector skin that moved that around. To restore the old view you will have to create an account, then click on "Torna al vecchio aspetto" on the side bar to go the the section of your user preferences where you can turn that off and restore the prior display. — xaosfluxTalk21:09, 2 July 2021 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Tech News
The next issue of Tech News will be sent out on 19 July.
Recent changes
AutoWikiBrowser is a tool to make repetitive tasks easier. It now uses JSON. Wikipedia:AutoWikiBrowser/CheckPage has moved to Wikipedia:AutoWikiBrowser/CheckPageJSON and Wikipedia:AutoWikiBrowser/Config. Wikipedia:AutoWikiBrowser/CheckPage/Version has moved to Wikipedia:AutoWikiBrowser/CheckPage/VersionJSON. The tool will eventually be configured on the wiki so that you don't have to wait until the new version to add templates or regular expression fixes. [40]
Problems
InternetArchiveBot helps saving online sources on some wikis. It adds them to Wayback Machine and links to them there. This is so they don't disappear if the page that was linked to is removed. It currently has a problem with linking to the wrong date when it moves pages from archive.is to web.archive.org. [41]
Changes later this week
The tool to find, add and remove templates will be updated. This is to make it easier to find and use the right templates. It will come to the first wikis on 7 July. It will come to more wikis later this year. [42][43]
There is no new MediaWiki version this week.
Future changes
Some Wikimedia wikis use Flagged Revisions or pending changes. It hides edits from new and unregistered accounts for readers until they have been patrolled. The auto review action in Flagged Revisions will no longer be logged. All old logs of auto-review will be removed. This is because it creates a lot of logs that are not very useful. [44]
@Matma Rex wrote mw:Extension:DiscussionTools/How it works. I understand that it contains the information you might need to get the Reply tool to work inside another script/tool. I post the link here in case anyone's interested.
notifications about new comments in any individual Section, regardless of whether the page is on your watchlist is a feature I would die for! Eagerly looking forward to that. -- RoySmith(talk)19:25, 29 June 2021 (UTC)
Certain scripts (Ohconfucius's formatgeneral, Sources, and Common Terms, as well as TemplateScript) only show up on pages with &action=submit appended to their URL, as well as some random text.
Are there any error messages in the browser console? Otherwise, the only thing I can think of would be to revert your common.js to a version where it worked and then add the scripts in question again one-by-one until you find the one that breaks it. – Rummskartoffel14:05, 25 June 2021 (UTC)
I get this:
Large output dump
load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:63 Uncaught TypeError: Cannot read property 'indexOf' of undefined
at Array.<anonymous> (<anonymous>:760:12)
at Object.ARA_Functions.getSuggestions (<anonymous>:380:24)
at Object.ARA_Functions.scan (<anonymous>:299:45)
at HTMLDocument.<anonymous> (<anonymous>:22:16)
at mightThrow (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:60)
at process (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:61)
mdlLogic.js:1 Uncaught URIError: URI malformed
at decodeURIComponent (<anonymous>)
at Function.t.EPPyTH (mdlLogic.js:1)
at t (mdlLogic.js:1)
at Object.mw.loader.load (<anonymous>:2:180)
at <anonymous>:6:11
at Object.<anonymous> (<anonymous>:1:724)
at mightThrow (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:60)
at process (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:61)
Uncaught TypeError: Cannot read property 'end' of null
at jQuery.fn.init.getCaretPosition (<anonymous>:106:580)
at jQuery.fn.init.$.fn.textSelection (<anonymous>:201:237)
at jQuery.fn.init.$.fn.wikiEditor (load.php?lang=en&modules=ext.wikiEditor&skin=vector&version=10s2l:11)
at Object.mw.addWikiEditor (load.php?lang=en&modules=ext.wikiEditor&skin=vector&version=10s2l:5)
at HTMLDocument.<anonymous> (load.php?lang=en&modules=ext.wikiEditor&skin=vector&version=10s2l:5)
at mightThrow (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:60)
at process (load.php?lang=en&modules=ext.CodeMirror|ext.CodeMirror.data|jquery%2Coojs-ui-core|jquery.ui&skin=vector&version=pg4sw:61)
Uncaught URIError: URI malformed
at decodeURIComponent (<anonymous>)
at Function.t.EPPyTH (mdlLogic.js:1)
at t (mdlLogic.js:1)
at Object.mw.loader.load (<anonymous>:2:180)
at Object.preloadDeflate (<anonymous>:158:106)
at load.php?lang=en&modules=ext.visualEditor.articleTarget%2Ccore%2CdesktopArticleTarget%2Cmwsave%2Cmwtransclusion&skin=vector&version=1ga9r:9
As was mentioned above, you should remove all of your scripts and then activate then one by one to find the one that is malfunctioning. If it is one you copied from someone else, they may be the best to help you with it. — xaosfluxTalk16:37, 29 June 2021 (UTC)
@Rummskartoffel: AFter testing this on my alt, I found that these pages load differently for an unknown reason which I couldn't locate. To clarify, this only appeared on the page which has the edit summary (show preview, show diff) on the same page as the editing window. (Or at least, before the big blue 'Publish changes' button is pressed.) ―Qwerfjkltalk19:17, 5 July 2021 (UTC)
I'm afraid I can't help you much more apart from repeating what I've already told you. It's going to come down to you finding, using the method I told you above, the script that causes the issue. That may be RedWarn, which does appear to run into some kind of problem, at least, or it may be any other script. Once you've found the problematic script, I or someone else can try and figure out why it breaks the others. – Rummskartoffel21:05, 5 July 2021 (UTC)
Do i exist?
Very minor question, but puzzling me nonetheless. In discussion at the current RfA someone mentioned XTools Admin Score; as one does, i took a look and, curious, entered mine own account. It says that the account is 0 days old. So i looked at Special:List Users, which also gives no creation date for LindsayH. Am i missing something very simple (most likely), or is there something wrong with my account? Thanks; happy days, LindsayHello19:06, 3 July 2021 (UTC)
It has the same for me, LindsayH. I recall seeing a mention somewhere that accounts created early on (before some sort of software change) can't currently have their ages computed. Sorry, I forget what the technical details are. Schazjmd(talk)19:21, 3 July 2021 (UTC)
Thank you. I did wonder if that might be it, but the same list page shows creation dates for LindsayJ, LindsayJo0307, and LindsayK.123, to take three at random, all of which were created prior to mine. Maybe it's something more "special" about you and me happy days, LindsayHello19:28, 3 July 2021 (UTC)
@LindsayH and Schazjmd: Accounts that were created before the user_registration field was added to the database (in December 2005) did not have that filed filled in. Where such accounts had edits, the timestamp of the first edit was used to fill in the field as a guess. But if you created your account before but did not make any edits until after, you won't have a registration date recorded. There's a bug open about filling the rest in (T20638), but it seems unlikely anyone will get around to it since they haven't for so long yet. Anomie⚔19:29, 3 July 2021 (UTC)
I second Schazimd's thanks; Anomie, what a fount of knowledge. I deliberately gave the section a provocative/silly heading; Redrose64, i'm so glad it provoked a good joke from you; happy days, LindsayHello05:24, 4 July 2021 (UTC)
I have a script or gadget that displays first and last edit dates on an editor's user page and I was wondering why some editors had no start dates. Learned something today! LizRead!Talk!00:37, 6 July 2021 (UTC)
Problem with River Lugg references
There seems to be a problem with some refs on the River Lugg article. I have used {{sfn |Jacklin |2015 |p=2}} twice, and Jacklin is defined in the bibliography. However, I am getting a "Cite error: The named reference "FOOTNOTEJacklin20152" was defined multiple times with different content (see the help page)." message. This is not normally a problem, since I have used {{sfn |Priestley |1831 |p=697}} twice in the same article, and that is ok. Any suggestions as to how to fix it? Thanks. Bob1960evens (talk) 09:40, 6 July 2021 (UTC)
I still don't get why valid names are required to use the dot notation. Is there even an ambiguous case where the parser wouldn't know if the user means the token or the table key? Alexiscoutinho (talk) 22:05, 6 July 2021 (UTC)
I doubt asking here is going to get you an answer about how parsing works at a basically theoretical level. Izno (talk) 23:25, 6 July 2021 (UTC)
Your link answers the question because it prominently shows: "The following keywords are reserved and may not be used as names". The reason for that rule would be buried in the details of how the Lua byte-code compiler works. Johnuniq (talk) 23:41, 6 July 2021 (UTC)
Should a broken tool, no longer being developed, be allowed to edit Wikipedia?
DAB Solver tool link was developed by User:Dispenser/Dab_solver. I saw "was" because I understand the tool is no longer being maintained. Yesterday it caused a storm in a teacup with an oversensitive editor who objected when I reverted a change that replaced a disambiguation link (not ideal) with a redirect (not idea) and instead I put the correct link in there. I also noticed that in the background, with the person using the tool being unaware, it is changing correct instances of {{Cite web}} (and its cite siblings) with {{cite web}} - despite the fact that other tools/bots change it from lowercase to uppercase. Given that the tool isn't being maintained, is not properly fixing dab links as it advertises, and is doing odd things in the background, should it be allowed to continue editing or be blocked? 10mmsocket (talk) 08:05, 6 July 2021 (UTC)
Just to note this does not seem to be happening only with Dab solver. It seems to be appearing whenever older-style cite parameter labels exist on a page. Whenever the page gets edited at all, the cite <param> labels are changed from Cite to cite. But actually isn't cite web, etc., the norm, anyway? I agree it is probably an unneeded, maybe annoying thing, but I do no think it's correct that other tools will then change it back? Surely that is not right?
If I use the ref tool, like so: [1] all the parameter labels in the template are already lower case. I guess I just don't understand what it's all about.
(I am the "teacup-oversensitive user" who was startled to be told they had "broken templates", with increasingly peremptory 2nd, 3rd, etc., "cease and desist" notices, before the first had barely had time to pop up for me. Lots of scary accusations and doom-laden orange bars, rather less explaining!) All over to you, the real WP-ers, now. Best, 49.177.30.125 (talk) 08:37, 6 July 2021 (UTC) Did I do the that mess above, somehow? 49.177.30.125 (talk) 08:40, 6 July 2021 (UTC)
While unmaintained Dab Solver works well. The tool doesn't select the links, the user does. So if sub-optimal links are being chosen that's not the fault of the tool. The other issues e.g. changing C to c in template titles can be avoided by unchecking the box that says "Apply common fixes" at the top of the tool. Nthep (talk) 13:06, 6 July 2021 (UTC)
Good point about the "apply common fixes". And again I take your point that the user needs to be more careful in choosing the right link - although the tool itself doesn't make any distinction between links to valid pages or links to redirects. I guess less experienced click-happy editors just get carried away and don't see the train wreck they're leaving in their wake. Fixing the tool would be best IMO, but at least I know now how to advise the inexperienced editors using it incorrectly. Many thanks. 10mmsocket (talk) 13:11, 6 July 2021 (UTC)
I don't know that this is the correct venue for this topic – not really 'technical' but more 'policy' I think. Still, tools that break stuff, especially those tools that are no-longer maintained (WP:REFILL is a prime example of that) should not be allowed to run. Yeah, I know, the editors who operate those tools are responsible for the edits that the tools make; but much too often, much too often, those tool operators accept whatever the tool suggests regardless of the results which leaves it to others to correct. I grow weary of cleaning up after such edits. I don't know how the IP editor created the citation above, but it is clearly wrong: {{!}} Thesaurus.com does not belong in |title= because that is not the title of the page that readers see. Been seeing more and more of these kinds of citations.
@Trappist the monk: the {{!}} Thesaurus.com titles most likely come from the User:V111P/js/WebRef script for generating cites from web pages. I know, because I sometimes use said tool, that it has a tendency to append the name of the publication in the title with that exclamation mark, even in cases where it also puts that publication name in the |work= parameter. I have no idea why, but I try to always remove it from the title myself. — Amakuru (talk) 13:07, 6 July 2021 (UTC)
Amakuru, Trappist: The tool isn't "appending" or "adding" the website name, it's obviously reading the <title> element from the target and using it unchanged for |title= instead of trying to figure out what the human reader sees as the title when viewing the page content. It's hardly mysterious, IMO. — JohnFromPinckney (talk / edits)14:38, 6 July 2021 (UTC)
Just because the tool is fetching information from <title>...</title> doesn't make that information correct. On the example page, for example, there are two <title>...</title> tags:
<titledata-react-helmet="true">OVERSENSITIVE Synonyms: 246 Synonyms & Antonyms for OVERSENSITIVE | Thesaurus.com</title> – source for browser tab
<title>Thesaurus.com</title>
and there is this which appears to be the source for |title= in the example above but doesn't match that text in the browser tab:
<meta data-react-helmet="true" property="og:title" content="Synonyms of oversensitive | Thesaurus.com"/>
Wrong information, no matter how easily used in a cs1|2 template is still wrong information. Permitting these tools to create malformed citations, like the one above, is a disservice to the encyclopedia and a burden imposed on other editors who have to cleanup the mess.
Its not malformed, it's slightly incorrect and easily fixed. More easily than a bare reference. Just fix the citation and move on. —TheDJ (talk • contribs) 21:33, 6 July 2021 (UTC)
No. It is malformed. Why should I have to fix, again, and again, and again, something that the tool-drivers created. The problem is that, all too often, the tool drivers (not just newbies but also some very very experienced editors) can't be bothered to fix what the tools create so it falls to others to make the fix. That is at best discourteous and at worst disruptive and detrimental to the encyclopedia. If I wrote a bot that performed as poorly as some of these tools perform, the community would rise up and shutdown the bot. Yes, I know, the tool drivers are (in theory at least) responsible for the edits that the tools they drive make. Alas, too many of them don't bother to fix the crap that is produced. I know this because I have fixed way too much of the junk.
Just on a point of order, 10mmsocket you are incorrect that it is "bad form to introduce redirects into articles". At most it's harmless, and in some cases it's preferable to have a redirect rather than an actual link to the article per MOS:RDR. Certainly the IP was doing good work and improving the encyclopedia by fixing links to dab pages, and a wet fish is due to you for attacking them for that. Cheers — Amakuru (talk) 13:14, 6 July 2021 (UTC)
I love fish, thanks. You do of course make a point, which I note. However, if the IP editor hadn't been overly sensitive this wouldn't have been an issue. While we all strive to be civi here, it is sometimes impossible not to offend some people. 10mmsocket (talk) 13:20, 6 July 2021 (UTC)
Every change to CS1|2 has the potential to break tools and bots. Since CS1|2 is changing all the time, any tool or bot not actively maintained should be suspended, the environment here is too dynamic to rely on users to fix mistakes manually. -- GreenC00:07, 7 July 2021 (UTC)
If any decision being made results in editors adding a reference with something like an extraneous template parameter vs not adding a reference (or even only adding a bareurl), I'm going to tend to support the former in most cases. — xaosfluxTalk09:43, 7 July 2021 (UTC)
Editing modes
Resolved
I've run into some errors, and after testing on my alt, it appears I'm opening edit pages differently. By default, th editing pages load with a toolbar at the top, but in my slt, it has features at the top, and the publish, show changes etc. buttons at the button. How do I get the latter editing mode as my default? (The latter editing mode also appears to have an older UI, and is similar to the Convenient Discussions one.) ―Qwerfjkltalk20:59, 6 July 2021 (UTC)
mw.hook('convenientDiscussions.commentsReady').add(function(){// comments_in_local_time.js import codeimportScript('User:Gary/comments in local time.js');// Backlink: [[User:Gary/comments in local time.js]]});
I think we've given you this advice before: unload ALL of your scripts and then try the one you want. When you have issues with someone else's personal script that you've imported the best place to start asking questions about it is on their talk page. As you are loading a very large number of personal scripts in User:Qwerfjkl/common.js it is certainly likely they will collide with each other, or that you are trying to use them outside the method their author intended them to be used with; even single syntax errors may cause everything in your script file to break. You may get another response here - but because you are choosing to use so many personal scripts at once, that is the risk you take. — xaosfluxTalk10:21, 6 July 2021 (UTC)
@Qwerfjkl: I didn't mean to come off harsh there, This just seems like it's approaching a "when I hit myself it hurts -- well stop hitting yourself" situation. A venue that may get you better support for this type of discussion is Wikipedia talk:User scripts, it is mostly attended to by other user-script enthusiasts such as yourself. If you have a general site-wide tech issue, or see a problem with one of the site-wide gadgets, VPT is the right spot though! — xaosfluxTalk10:28, 6 July 2021 (UTC)
@Qwerfjkl: Xaosflux is exactly correct IMO; when you have this many scripts, it's not really reasonable for us to debug everything for you. That said, I'm guessing the reason this script isn't executing is that I don't see any hook named "convenientDiscussions.commentsReady" in User:Jack who built the house/convenientDiscussions.js, which is presumably where you're expecting that hook to come from. So you're attaching the script in a handler to a hook that never seems to be fired. Writ Keeper⚇♔13:49, 6 July 2021 (UTC)
Along with the things already mentioned (~200 scripts? How can you find the actual article text?), I noticed somewhat by chance, that you even have duplicated entries. Lines 46 and 48 appear to be identical (replacing the same script commented out at line 12), although there's no benefit in installing it twice. I'm rather amazed that any scripts work in that huge pile, not that the last one installed doesn't. — JohnFromPinckney (talk / edits)14:24, 6 July 2021 (UTC)
I have the same snippet in my common.js, except that it's using User:SD0001/comments in local time.js instead of the Gary version – I think this fork is better compatible with Conv Discussions, but the Gary version was also working for me so not sure if this would solve the issue for you. – SD0001 (talk) 15:09, 6 July 2021 (UTC)
@SD0001 Thanks, using your script worked! I've also cleaned up my common.js page a bit. I only asked here in case I'd misunderstood how to load the comments in local time script. Sorry for not making that clear. ―Qwerfjkltalk15:48, 6 July 2021 (UTC)
As another CD-related error, this:
o/w/api.php?titles=r&origin=*&format=json&formatversion=2&uselang=content&maxage=86400&smaxage=86400&action=query&prop=revisions|info&rvprop=content&rvlimit=1:1 Failed to load resource: net::ERR_NAME_NOT_RESOLVED
has been fulling up my console (reaching over 100 messages), and also preventing me from viewing messages posted with CD. ―Qwerfjkltalk15:54, 6 July 2021 (UTC)
Can you provide more details about this error? It doesn't seem to come from Convenient Discussions.> preventing me from viewing messages posted with CD Can you elaborate? Jack who built the house (talk) 17:02, 6 July 2021 (UTC)
@Jack who built the house to clarify, once I click Reply, it doesn't show the comment as posted, even though it has. It appears as if I have done nothing (after loading a bit) and reclicking the button can post a comment multiply times. An error message saying check your console also appears. ―Qwerfjkltalk18:57, 6 July 2021 (UTC)
I think that's because the path "o/" doesn't exist on enwiki (unless one of the scripts uses other server). And it probably comes from that convoluted script in 2nd line of your common.js. MarMi wiki (talk) 18:51, 6 July 2021 (UTC)