Wikipedia:Village pump (technical)/Archive 36

• Village pump
  • Policy
  • Technical
  • Proposals (persistent)
  • Idea lab
  • WMF
  • Miscellaneous

Village pump (technical) archive  

This page contains discussions that have been archived from Village pump (technical).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.

---

< Older discussions · Archives: A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, Z, AA, AB, AC, AD, AE, AF, AG, AH, AI, AJ, AK, AL, AM, AN, AO, AP, AQ, AR, AS, AT, AU, AV, AW, AX · 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217

Backwards edittime?

Am I the only one who finds this backwards? ♫ Melodia Chaconne ♫ (talk) 17:10, 19 April 2008 (UTC)

Happens for me too - how bizarre. Happy‑melon 17:39, 19 April 2008 (UTC)

I took a look. Based on the text people added the revision numbering is in the right order but the time stamps are in the wrong order. Seems some server has a faulty clock. The function to compare "older/newer edit" shows them in the right order (revision number order). But apparently the history view sorts them in time order. Revision numbers like that must be globally coordinated between servers, but I guess they only synchronize the clocks every now and then. I hope that server will soon be taken off-line, otherwise we will get very funny history lists if that clock keeps drifting even more.

--David Göthberg (talk) 17:46, 19 April 2008 (UTC)

Y Done. I reported it and the dev Mark fixed it. He said: "Server 133's time was off."

--David Göthberg (talk) 18:20, 19 April 2008 (UTC)

Seems to have been fixed: mark: srv133's time was off, corrected.

I rather enjoyed thinking I'd reverted vandalism before it occurred. --jpgordon^∇∆∇∆ 19:12, 19 April 2008 (UTC)

Is the ordering in history going to be fixed? --Random832 (contribs) 21:41, 19 April 2008 (UTC)

My guess is that the server admins will not search through and fix the database since they are too busy. Normal Wikipedia admins can not change such time stamps. I think we will have to live with those erroneous timestamps in the history lists.

--David Göthberg (talk) 22:22, 19 April 2008 (UTC)

Does this problem concern the order, or are the times altered as well? I made two subsequent edits to the Policy section of the Village Pump today; the first was at 12:30 UTC and the second three minutes; instead of 12:33, however, it was registered as 12:13. This diff makes it look as if I deleted my own message. If one goes an edit back, it will all appear normal. If one then goes an edit forward, the diff will look just fine (but still with an erroneous time, which has not yet been fixed). Weird... Waltham, The Duke of 23:06, 19 April 2008 (UTC)

But they could fix the history to order by revision id. Except that'll mess up for old deleted revisions. Maybe have an option? --Random832 (contribs) 00:55, 20 April 2008 (UTC)

It's been much worse before. Making the revID fields and dates in synch would be nice but I'd imagine it would require a big database schema change so that old diff links aren't broken in the process. Graham87 05:37, 20 April 2008 (UTC)

Those diffs are indeed troubling. I haven't run into anything like those — yet! Gary King (talk) 06:46, 20 April 2008 (UTC)

My perferences → Math → Recommended for modern browsers

What does this option actually mean? I see no difference with "HTML if possible or else PNG" If there's really no difference, why there's such an extra option? --Quest for Truth (talk) 14:37, 20 April 2008 (UTC)

The Modern Browsers option shows PNG first, and if not possible, then HTML; the other option, HTML or else, PNG, is the opposite. Gary King (talk) 19:04, 20 April 2008 (UTC)

It is asking whether it should render everything in HTML, everything in PNG, or work out which is more efficient and do that. asenine ^t/_c 03:39, 21 April 2008 (UTC)

show only minor edits

Looking at recent changes I can select to hide minor edits, or to see minOr and non-minor edits together. I don't seem to be able to list only minor edits. Is this possible? Is anyone checking to see that minor edits actually are minor and not non-minor in "disguise"? Dan Beale-Cocks 21:27, 20 April 2008 (UTC)

You can't show only minor edits, as far as I know. There might be a script out there that does this, though. Gary King (talk) 00:57, 21 April 2008 (UTC)

Extension:SpamRegex

A MediaWiki extension that may be helpful to us is the SpamRegex extension. We currently have the Spam-blacklist, but this only prevents direct links from being saved. It doesn't prevent URLs wrapped in <nowiki> tags or URLs put in edit summaries. Depending on how the regexes are written, it could also make combating spam slightly more difficult (harder to discuss) so it would only be used/needed in the more extreme situations. Just the edit summary blocking alone may make installation worth it. I asked Brion about installing it and he said it should be fine, so I'm proposing it here for community review before requesting it. Mr.Z-man 03:42, 17 April 2008 (UTC)

Seems like a great idea to have on en.wiki. Fully support. In fact, perhaps all of the blacklists (title, username, spam, filename prefix) could be merged into a single extension. --MZMcBride (talk) 03:46, 17 April 2008 (UTC)

Sounds like a very good idea to me, too, so long is it isn't too CPU or memory hungry. SQL^{Query me!} 04:38, 17 April 2008 (UTC)

Sounds yummy. --slakr^\ talk / 05:30, 17 April 2008 (UTC)

this would make fighting spam a hell of a lot easier. β^{_{command 2}} 19:33, 17 April 2008 (UTC)

Seconded, if this could be installed as soon as possible, preferably across all the sites - it should make dealing with spamming on the smaller Wikis much easier to deal with (not that it's something I've had much time to deal with recently). Nick (talk) 22:07, 17 April 2008 (UTC)

Request made. Happy‑melon 18:02, 21 April 2008 (UTC)

Preventing loss of edits when article goes into lock for maintenance mode

I just lost a few edits when I tried to save an article after editing. I backed up a page but no edits.

Worse, the system suggests that you return to the article which will really cause a loss of the edited page.

First, I don't think that the system should suggest to an editor that s/he return to the article. The editor is knowledgeable enough to navigate their way around. Giving them a sure fire way of losing their edits is probably not a good idea!

Second, it would be better to somehow buffer the edits so they can be made later as happens when there is an edit conflict.

Thanks.

In most instances, the reason the database locks is because the slave servers fall too far behind the master server: essentially, people are making too many changes (edits, deletions, recaches, etc) for the system to keep up. It locks the database to stop the mayhem until it's cleared out enough of the backlog to be stable. Adding requests to a queue would have the exact opposite effect, making the problem worse rather than better. If you don't want to lose your edit text, just use a browser other than IE - most sensible browsers automatically retain the contents of fields on previous pages, so if you click 'back', your edit will still be there. Happy‑melon 16:16, 19 April 2008 (UTC)

If you're using IE, refreshing the page and answering "yes" to the question about resubmitting information will resubmit the edit. Just do that until it eventually comes through. I suppose that would work on other browsers and be easier than hitting the back button. Graham87 05:25, 20 April 2008 (UTC)

The first commenter here is right. We should change the message that people get during database locks. That is, we should instead advice them how to not loose their edited data in the different browsers. (Back button in Firefox, refresh/reload in IE with answering yes to resubmitting the data, and so on.) But of course, we should also ask them to wait 30 seconds or more before they do it to let the servers catch up.

--David Göthberg (talk) 10:52, 20 April 2008 (UTC)

This should just re-display the form with all the submitted data, like a preview. That's the standard solution to this sort of thing. —Simetrical (talk • contribs) 02:40, 22 April 2008 (UTC)

Broken links to old image versions

For some reason, the links to old versions of Image:Ipod backlight.jpg ([1] and [2]) give a 404 error. Any idea what's wrong? —Ilmari Karonen (talk) 17:34, 19 April 2008 (UTC)

The files are present on a different server... I'm not sure what's wrong exactly. Very strange, will try to investigate. --brion (talk) 22:26, 21 April 2008 (UTC)

Special pages

• Special:SpecialPages

Is there any particular reason that the new "sections" (which are very nice, by the way), are in their current order? And also, any reason why we have these sections and not some others? - jc37 23:09, 19 April 2008 (UTC)

I believe that this is hardcoded into the MediaWiki software (that happened in rev:33197). This means the order, number of sections, which section certain special pages are in, etc. can't be edited by sysops via the MediaWiki namespace. However, the titles of the section can be edited at MediaWiki:Specialpages-group-maintenance, MediaWiki:Specialpages-group-other, MediaWiki:Specialpages-group-login, MediaWiki:Specialpages-group-changes, MediaWiki:Specialpages-group-media, MediaWiki:Specialpages-group-users, MediaWiki:Specialpages-group-needy, MediaWiki:Specialpages-group-highuse, and MediaWiki:Specialpages-group-permissions. Also, a code similar to the following (copy and pasted from here) can be added to MediaWiki:Common.css (or a skin-specific .css or a user's .css subpage) to change the look of it slightly I think. Aside from that, no changes are possible unless you are a dev. FunPika 00:09, 20 April 2008 (UTC)

/* Special:SpecialPages styling */
h3.mw-specialpagesgroup {
	background-color: #dcdcdc;
	padding: 2px;
	margin: .3em 0em 0em 0em;
}

table.mw-specialpages-table {
	background-color: #f9f9f9;
}

Actually...the CSS part didn't work right when I tried it in my modern.css and my common.css subpages. FunPika 00:26, 20 April 2008 (UTC)

I find the categories to be much more useful than before, when it was all in one alphabetized list. The developers made a smart move on this one! Gary King (talk) 06:44, 20 April 2008 (UTC)

A first draft of this had a system message that allowed the categorizations to be edited, but that was viewed as overly complicated. It would lead to unnecessary inconsistency between sites, and maintenance problems (when devs add a new special page, how does it get added to the customized local message?), for relatively little benefit. So the new version has the organization programmed into the software, yeah, basically according to what Aaron thought was a good arrangement. —Simetrical (talk • contribs) 02:46, 22 April 2008 (UTC)

Searching for discographies from Google, on Wikipedia, does not return correct articles?

If I'm on Google and search only Wikipedia (meaning using site:en.wiki.x.io search term) and I search for, say, 'Beatles discography', I don't get the Beatles discography list to return as a result. What's up with that? I've always wondered about that. Gary King (talk) 06:49, 20 April 2008 (UTC)

I don't know, but searching Wikipedia for Beatles discography (press 'search', not 'go'), as with all pages, doesn't bring up the page with that name as the top result. Could they be related? ...... Dendodge.Talk^Help 10:13, 20 April 2008 (UTC)

You may find the "allintitle" keyword useful [3]. Probably The Beatles main article is orders of magnitude more popular than the discography, or at least enough to make the exact title match less favored (shrug). — CharlotteWebb 11:27, 20 April 2008 (UTC)

I tried it, and as usual with Google, I got "about 3,860" results. So I'm sure it's there somewhere! Looking at the results on page 1, I see numerous entries that make reference to the page you're looking for, for example, the 3rd result is for "George Martin" and highlights "Main article: The Beatles discography" so you should have no trouble getting there from here. I suppose the answer to your question is, it depends on how the article is worded. It does not have the words "Beatles" and "discography" side by side anywhere within the text. Perhaps the articles on page 1 of the search have these 2 words together more than once, giving them precedence. I also notice that none of Google's page 1 results have the 2 words as part of the page title; you would think that has precedence, but apparently not. --A Knight Who Says Ni (talk) 14:19, 20 April 2008 (UTC)

Huh? Clicking on this link [4] I get exactly two (2) results, which are The Beatles discography and Beatles discography, one of which is a redirect to the other. — CharlotteWebb 02:03, 22 April 2008 (UTC)

Hmmm, it seems very odd ... the page title is a very important part of Google's ranking system, and a search for "beatles discography site:en.wiki.x.io" should bring it up as the first result. Perhaps there was a server glitch at the time the Google bot was crawling the site? The article The Beatles discography was never deleted or moved to a whacky title according to its log. When I checked out this problem earlier, I thought that Google might have blocked the title because it contained the letters "disc", which is part of "discussion"; I was thinking about Google's voluntary blocking of all talk pages in search results. However, a search for "discography" disproves that theory. The link to the article from this page will probably force Google to include it in their index, because it often patrols frequently updated pages like this one. Graham87 15:56, 20 April 2008 (UTC)

First of all, I'm glad that I'm not the only one experiencing this 'issue'. Secondly, this has been happening for months, at the very least; I've only brought it up today. It's something that has been bugging me for a long time, because logic says that Google is smart enough to know that the page is about Beatles discography. They have their algorithm locked down, so I think the problem lies in Wikipedia's robot.txt or another configuration found at Wikipedia. The problem does not lie only with Beatles discography, but really with any discography article here. I think it even happens to filmographies. Another reason that this is happening might be because Google does not consider discographies nearly as important as the main article, but if that's the case, I would really disagree, especially if someone explicitly searches for 'discography' in their search term. Gary King (talk) 18:16, 20 April 2008 (UTC)

The Beatles discography is currently the 67th hit on "Beatles discography" site:en.wiki.x.io for me. I don't know why it's so low but it is there, cached by Google on 18 April. The 68th hit is the same article with the redirecting (in the Wikipedia sense) url http://en.wiki.x.io/?title=Beatles_discography. Could this copy registered by Google lower the ranking? Maybe http://en.wiki.x.io/robots.txt should exclude http://en.wiki.x.io/?title PrimeHunter (talk) 22:39, 20 April 2008 (UTC)

New pages by user

I've noticed that quite recently from the New Pages screen that the box below namespace where you could enter a user's name has gone. Why is this and can it be restored? Lugnuts (talk) 10:27, 20 April 2008 (UTC)

It must have been a recent change to MediaWiki. It's strange that that would be removed, but on the other hand, I've never noticed that feature and therefore probably did not find it that useful... Gary King (talk) 19:10, 20 April 2008 (UTC)

There are some missing needed indexes on four wikis. Rather than have some live hacks floating around, the whole thing was removed from all wikis for now. It will be re-enabled when the SQL indexes are added. Voice-of-All 23:14, 20 April 2008 (UTC)

Thanks for the replies. Do you know when the SQL indexes will be added? Lugnuts (talk) 11:14, 21 April 2008 (UTC)

History

When we pick say 2 versions and have 4 intermediate revisions not shown, there's only an arrow that points to the older version and an arrow that points to the new version. I would be helpful if we could have arrows for THE intermediate versions, both for the newer and the older. Please post this on the bugzilla thing or whatever because I don't have an account. Thanks, I really appreicate it!!68.148.164.166 (talk) 11:48, 20 April 2008 (UTC)

If you want to see what changes have been made for each intermediate version, all in one view, then that's not possible and I don't think will happen for some time. One of the big reasons is because if you compare two versions that have 50 intermediate versions, then it can get messy with arrows, very quickly. It's also more processor-intensive. Gary King (talk) 19:06, 20 April 2008 (UTC)

I think you misunderstood the post. It's about which diffs to link and not about displaying more than one diff at a time. This diff says "4 intermediate revisions not shown", and has the link "Older edit" with a left arrow and "Newer edit" with a right arrow. Suppose we number revisions 1,2,3,4,5,6,7,8 and the diff is for (2,7). Then "Older edit" is (1,2) and "Newer edit" is (7,8). The poster wants two more links, for (2,3) and (6,7), so you can go in both directions from each revision in the original diff. I have also missed that ability. PrimeHunter (talk) 12:45, 21 April 2008 (UTC)

Personal toolbar jumpig to the left

I suddenly notice that when I hoover over my personal toolbar, it jumps to the left side of the screen. I've noticed it on Commons a few months ago. But now I cannot see any changes to common or monobook files that coudl cause this. Anyone else seeing this? Using IE6. ~~—Preceding unsigned comment added by Edokter (talk • contribs) 13:25, 20 April 2008 (UTC)

I used to have the same problem intermittently with IE6 & IE7, I don't recall having it with Safari on WinXP, which I use now. DuncanHill (talk) 16:09, 20 April 2008 (UTC)

Might be something you added to your monobook.js recently? Gary King (talk) 19:04, 20 April 2008 (UTC)

Nah, must be something with the computer at work; problem went away when I got home. — Edokter • Talk • 19:12, 20 April 2008 (UTC)

This is an old IE bug. Upgrade to IE7, then bug Microsoft if it still does it. --brion (talk) 22:10, 21 April 2008 (UTC)

Date and time in preferences.

The formatting has

“ <br><small>This will adjust the times shown on your watchlist and recent changes, but not in signatures on talk pages, which will always be in UTC.</small> ”

in plain text. You might want to fix that. asenine ^t/_c 03:37, 21 April 2008 (UTC)

That's very strange, since MediaWiki:Timezonetext has not been edited since December 2006. I've removed the markup, so at least users won't be quite so confused about it. —Remember the dot ^(talk) 04:14, 21 April 2008 (UTC)

Seems to have been caused by r33495. It should be easily fixable. —Simetrical (talk • contribs) 02:56, 22 April 2008 (UTC)

Allowing editors to create a toolbar of quick links

There's not a whole lot of space available on the WP page, but I'd still like to be able to add some personal, user chosen links. Next to my username, for example, there's space to add another 5-6 links. Many functions on WP require 3-4 moves to get to (Articles for Deletion, for example). I could add the AfD page to my Firefox toolbar, but it's already full of useful links. Also, there's some space under the Toolbox for more links. OptimistBen | talk - contribs 04:23, 21 April 2008 (UTC)

This is pretty easy in javascript using addPortletLink();, for example

addOnloadHook(function() {
  addPortletLink('p-tb','/wiki/Somewikilink','Some wikilink');
  addPortletLink('p-personal','/wiki/Somewikilink','Some wikilink');
  addPortletLink('p-navigation','/wiki/Somewikilink','Some wikilink');
  addPortletLink('p-cactions','/wiki/Somewikilink','Some wikilink');
});

Just add that to your Special:Mypage/monobook.js and hit 'preview'. Play around with it (but use preview when possible, much easier than saving and reloading cache each time). More detailed instructions can be found in wikibits (scroll down or page search for 'addportletlink'). --Splarka (rant) 08:10, 21 April 2008 (UTC)

browsing by letter in categories

Category pages used to have a feature for browsing by letter, but I can't find it anymore. Was this feature removed for some reason? --Ixfd64 (talk) 08:09, 21 April 2008 (UTC)

It's actually a template: {{categoryTOC}}. It's not present on all categories, but you can add it to any that you think need it. Happy‑melon 08:59, 21 April 2008 (UTC)

Thanks! This template is very helpful. MediaWiki should include this functionality as a built-in feature! --Ixfd64 (talk) 09:02, 21 April 2008 (UTC)

Could be. It's only helpful on large categories, though, and it assumes that pages are named in Latin, which isn't necessarily the case. It would require some thought to do properly in the software. —Simetrical (talk • contribs) 02:58, 22 April 2008 (UTC)

There does seem to be a problem, though. Category:Stub categories has this template, which doesn't show up! Is anyone else having this problem? --Ixfd64 (talk) 09:05, 21 April 2008 (UTC)

It shows up fine for me. PrimeHunter (talk) 12:29, 21 April 2008 (UTC)

OK for me. --— Gadget850 (Ed) ^talk - 12:38, 21 April 2008 (UTC)

It's probably a browser problem. I just did a test - it works on Firefox but not Internet Explorer. --Ixfd64 (talk) 19:55, 21 April 2008 (UTC)

Sorry

Sorry, I'm sure this has probably been asked a million times, but how do you archive your own talk page? I keep deleting mine when it gets too long, but I'm pretty sure that is not the right way to go about it...!--seahamlass 11:53, 21 April 2008 (UTC)

See Help:Archiving a talk page. --— Gadget850 (Ed) ^talk - 11:59, 21 April 2008 (UTC)

Transclusion within a transclusion doesn't update

This is an issue I've noticed with my userpage, so I'll admit it's not really that pressing, but I could see how it may affect the mainspace if there are instances of transclusions within transclusions out there. Anyhow, User:Xenocidic/infobox/status is transcluded on User:Xenocidic/infobox which itself is transcluded onto my userpage, but the status doesn't seem to update properly when viewing the userpage... Is it too deep a recursion? Thanks in advance. xenocidic (talk) 12:44, 21 April 2008 (UTC)

It's just a caching issue: the appearance of your userpage is cached by the servers to improve performance. When you update your status indicator, the system notes that all the pages on which that template is trancluded need to be recached. So they're put into the job queue... which is currently about 8 million articles long. If you left everything alone, your page would eventually be recached... in a week or so. To force an immediate recache and ensure your userpage displays correctly, you just need to do a null edit to your userpage after you update the template. Happy‑melon 13:54, 21 April 2008 (UTC)

Right, I had figured out that workaround... I guess it's sub-optimal but necessary. Thanks for the response! edit: oh, I see that a null edit is actually different than a dummy edit, which I was doing. that's even better. thank you again. xenocidic (talk) 13:57, 21 April 2008 (UTC)

Patrolled is not displayed for anons

I noticed that anons do not see yellow color on Special:NewPages anymore. Is this permanent and was this really a good change? If any dev reads this, I would like to remind you that some other projects have a separate patroller user group, which means this change affects most logged in users. P.S. Enwiki admins might want to fix MediaWiki:Newpages-summary. —AlexSm 19:50, 21 April 2008 (UTC)

WhatLinksHere problem

Is there a problem with Special:WhatLinksHere - when selecting the article namespace it now returns all links from all namespaces. Keith D (talk) 23:50, 21 April 2008 (UTC)

I have the same problem. When selecting the article namespace and clicking Go, the namespace selector changes to all, and links from all namespaces are shown. Every other namespace worked in my test. PrimeHunter (talk) 00:01, 22 April 2008 (UTC)

This should be working now; there was an error in the input validation. --brion (talk) 01:19, 22 April 2008 (UTC)

(hist) (diff) vs. (diff) (hist)

Why is the layout on the Watchlist page (diff) then (history) but on the Contributions page (history) then (diff) ? TheRedPenOfDoom (talk) 04:51, 22 April 2008 (UTC)

Good question. I'd suggest filing a bug report asking that the Contributions page use (diff) (hist) like Special:Watchlist and Special:Recentchanges. If you file a bug, tell me what bug number it is and I'll speak up for the change too. —Remember the dot ^(talk) 05:06, 22 April 2008 (UTC)

Also, the timestamps are moved as I recall. Standardizing all of it would be nice. --MZMcBride (talk) 05:09, 22 April 2008 (UTC)

Archive BOT modification proposal

Hi there. To anyone with the skills interested, there's a tweak that might help make our lives easier with the archiving bots:

Problem: Usually, when we link to a discussion from, say, WP:AN we do it in the simple form of "Please see [[WP:AN#Subheading XYZ]]". Then, a bot comes along and moves this XYZ subheading to an archive (numbered as archiveABC), and those links are lost, so anyone interested to go back to that discussion will have to search using the datestamps (if they exist).

Solution: Probably there can be some sort of tweak in the archiving BOTs: Although I'm not very familiar with the way these things work, I assume they can first see Special/WhatLinksHere for the page they're archiving, then go the respective pages, and tweak the comments by adding the respective archiveABC to which they sent the particular subheading. This way the comment will link to the newly created archive page.

I was wondering if I should post this directly to Miszabot or any of the others, but I think it should be agreed whether this is a necessary feature first. Feel free to move my comment if inapplicable for this venue. Thoughts? NikoSilver 13:56, 18 April 2008 (UTC)

My take is it would unnecessarily add a lot of work. These get fixed manually when they are important. 199.125.109.88 (talk) 14:01, 18 April 2008 (UTC)

Just to add, only for Wikipedia:Administrator's noticeboard there are more than 4000 links [5], and there are another c.7000 for WP:ANI so far.[6] I imagine that WP:3RR and the rest will have links of a similar order of magnitude. (not saying if this is a lot or not, nor implying that all those links are not current - but e.g. AN's last archive has less than a dozen [7] -if one excludes the numerous links from the archive template existing in all AN and ANI archives). NikoSilver 14:29, 18 April 2008 (UTC)

Sections on ANI are archived after no one has edited them for a certain time. Presumably this means that either the situation is resolved or no one cares anymore. In either case, there's not much of a reason to update the links. Mr.Z-man 17:31, 18 April 2008 (UTC)

User:ClueBot III does this. It does not, however, support numbered archives, only dated archives. -- Cobi^(t|c|b) 05:09, 19 April 2008 (UTC)

Werdnabot indexes all the sections it archives in a nice list, so it's fairly easy to find them. — Werdna talk 14:14, 22 April 2008 (UTC)

@Z: I've tried to follow past discussions (including WP:RFCs, WP:RFC/Us, even saw "bad" links in WP:ARBCOM cases) and I was lost several times already. NikoSilver 00:53, 23 April 2008 (UTC)

@W: Index helps, but what if you don't remember the section title (as usual) and/or the archive has 187 sections (as usual)? Doesn't this limit the valuable content of a previous discussion to the connoisseurs only? NikoSilver 00:53, 23 April 2008 (UTC)

@C: I guess I'll have to wait until Cobi materializes his promise. :-) Thanks. NikoSilver 00:53, 23 April 2008 (UTC)

@All: When he does, wouldn't it be super if we could archive those high-traffic pages with his bot? NikoSilver 00:53, 23 April 2008 (UTC)

Apostrophe in ifeq

Can't get Template:Episode list/McLeod's Daughters to work, and I believe it's because of the apostrophe in the article title's name. The template is a hack of sorts, it allows editors to transclude the episode list on season pages onto the main episode list, but hiding the ShortSummary field to save space (if anyone is wondering, we're also looking for a way to make this work generically, so we don't have to make individual sub-templates for each show).

It works like this:

{{#ifeq:{{PAGENAME}}|List of McLeod's Daughters episodes|

Then it lists the template minus the ShortSummary parameter when the template is viewed on List of McLeod's Daughters episodes (when true), but includes it on individual season pages like McLeod's Daughters (season 1) (when false). However, it's showing the ShortSummary on both the season pages and on the main episode list. -- Ned Scott 19:16, 22 April 2008 (UTC)

Fixed, I think. You need to play around with parameter defaults, but the #ifeq statement now works. Happy‑melon 21:17, 22 April 2008 (UTC)

Seems like there might be a bug here. {{PAGENAME}} in the ifeq function is being escaped - ' becomes &#39; and " becomes &quot; . Is this supposed to happen? --- RockMFR 21:25, 22 April 2008 (UTC)

I shouldn't think so - surely that should only happen with {{PAGENAMEE}}. You can't use active wikimarkup in pagenames anyway, and if you're doing things where live punctuation is a problem, you'd use the URL-safe version - I don't think things in PAGENAME should be escaped. Am I missing something? Happy‑melon 21:30, 22 April 2008 (UTC)

Ahh, it's already reported - bugzilla:12448. --- RockMFR 21:38, 22 April 2008 (UTC)

Invalid email confirmation

A problem was reported at Help talk:Email confirmation#Invalid confirmation code. The code may have expired. If an unconfirmed email address is registered then Special:Preferences has a link called "Confirm your e-mail address". Clicking this link posts a confirmation code that did not work in any attempt by me and another user. If you change mail address and save then a code is automatically posted to the new address. This automatically posted code has worked fine in every attempt. Apparently the two ways to get a confirmation code are treated differently and one of them is buggy. PrimeHunter (talk) 23:32, 22 April 2008 (UTC)

Should be fixed now. A bug was introduced last week or so due to some internal code changes; when called that way, the new token value wasn't getting saved. --brion (talk) 23:48, 22 April 2008 (UTC)

Thanks for the fast response. It works fine for me now. PrimeHunter (talk) 00:12, 23 April 2008 (UTC)

Replacing "new section" with "+"

There was previously a "gadget" in Special:Preferences that allowed the "new section" tab to appear as "+" ... has it been disabled? Black Falcon ^(Talk) 00:47, 23 April 2008 (UTC)

I still have it and it works fine. It's under the "Gadgets" tab with the text: Change the "new section" tab text to instead display the much narrower "+". PrimeHunter (talk) 01:14, 23 April 2008 (UTC)

User:Cacycle disabled it without discussion, on the theory that it was an undiscussed addition (despite the discussion on WP:VPR) because the proper forms had not been filed. --Random832 (contribs) 01:18, 23 April 2008 (UTC)

Will we still have the option for "+" at the end of it all? It still works for me and I still want it! Franamax (talk) 01:32, 23 April 2008 (UTC)

OK, it's back... Black Falcon ^(Talk) 01:36, 23 April 2008 (UTC)

This should be discussed at Wikipedia:Gadget/proposals, not here. Сасусlе 03:19, 23 April 2008 (UTC)

Mediawiki non-existing

Is this supposed to happen? When I use search from within en:wiki for "mediawiki:download" I get a page that says "download", and when I search for "mediawiki:subversion", I get a titled but empty-boxed page. This may be my first attempted search in MediaWiki space, is it supposed to give me back a totally uninformative result? Franamax (talk) 01:39, 23 April 2008 (UTC)

These are labels used within the MediaWiki software. Examples: MediaWiki:Talk is the label for the talk page; MediaWiki:Addsection is the label for the tab at the top of the page to create a new section on a talk page; MediaWiki:Move is the label for the tab to move a page. MediaWiki:Subversion does not exist which is why you get a tab to create the page. --— Gadget850 (Ed) ^talk - 03:14, 23 April 2008 (UTC)

Special:Newpages by users

What happened to the possibility of obtaining a list of pages recently created by a user? Circeus (talk) 02:39, 23 April 2008 (UTC)

Is this the same issue as discussed just above? Franamax (talk) 03:18, 23 April 2008 (UTC)

Yup. I did not think of adding a space in my text search. *sigh* Circeus (talk) 03:44, 23 April 2008 (UTC)

Gadget discussion

In response to recent discussions on this page and Wikipedia:Village pump (proposals) several user preferences gadgets have been created. Some of them were deployed but were today removed by one editor.

The gadgets are:

• "Addsection +" – Adds back the "+" instead of the "new section" in the middle talk page top tab.
• "Diff underlining or borders" – Adds a red dotted border or red or yellow-green underlining to text that has changed in diffs. Makes it possible to see where spaces have been added or removed and makes it easier to find small changes like changes in punctuation.
• "Tighter page top tabs in MonoBook" – Gives the page top tabs tighter margins and paddings so they take up less space. Good for users with many extra tabs and/or low screen resolution.

I would like to call attention to the discussion of these gadgets at Wikipedia:Gadget/proposals#Tighter page top tabs in MonoBook (and down from that point in that page). The main question is: Should these gadgets be added as options in the user preferences or not?

--David Göthberg (talk) 06:28, 23 April 2008 (UTC)

Coordinates are overwriting the top line

This was brought up on 27 March 2008 but not answered and is still a problem. I believe from earlier discussion that the coordinates are located a fixed number of pixels from the top and end up being right over the Article name when there is a fund appeal. Did someone tweak the number of pixels recently and get it wrong? 199.125.109.88 (talk) 22:15, 9 April 2008 (UTC)

I noticed that recently as well. It's rather annoying. -mattbuck (Talk) 12:30, 10 April 2008 (UTC)

This can be fixed with the following CSS:

#coordinates {
  top:4.5em;
}

Just place that in your monobook.css. — Bob • (talk) • 04:28, 11 April 2008 (UTC)

Thanks, but IPusers don't have monobook.css pages. Howabout just fixing the glitch? 199.125.109.104 (talk) 17:15, 11 April 2008 (UTC)

I would if I could. You can see if someone else is willing to make the change, elsewise I suggest creating an account if it bugs you that much. — Bob • (talk) • 22:55, 11 April 2008 (UTC)

The problem is that no one has figured out so far WHY this suddenly started happening. Without understanding the problem people are a bit reserved about fixing it. --TheDJ (talk • contribs) 23:04, 11 April 2008 (UTC)

I think the horizontal rule, the line, got moved down, or I guess that would be up, no down, no up, no down making them collide. Maybe the size of the graphic at the left got changed, or something else that contributes to the location of the horizontal rule? 199.125.109.88 (talk) 02:30, 12 April 2008 (UTC)

Somehow the space allowed for the page title seems to be wider than it used to be, so the bar is lower than it used to be before. There must be some other images of Wikipedia pages that could be looked at to count pixels. 199.125.109.88 (talk) 02:39, 12 April 2008 (UTC)

Yikes. 743 of them in no particular order at Category:Screenshots of Wikipedia. The width of the space where the page title goes seems to be the culprit though. 199.125.109.88 (talk) 02:53, 12 April 2008 (UTC)

• OK there are 2 issues:
  1. When the sitenotice is present, it messes up the alignment for coordinates
  2. When coord is included from a table that has "font-size: 90%" for instance, then this value is inherited by the coordinates element. On this fontsize the top alignment is based (3.5em) An em is directly proportional to fontsize, so the offset of the top will be too small and the element will not be "low" enough. I'm looking into some ideas that could be used to solve this issue. However, this is not a new problem really and it seems that almost every potential solution has its drawbacks. --TheDJ (talk • contribs) 19:37, 12 April 2008 (UTC)

As I feared there is little that can be done to improve this. If we use monobook, we want to use that specific location and not rewrite the main monobook skin or the way coordinates are included, the only way to properly fix this is to use Javascript to move the DOM HTMLelement. See template talk page for an example of how that would work. --TheDJ (talk • contribs) 23:03, 12 April 2008 (UTC)

I'm confused. You've got about a centimeter of blank space to drop the coordinates into and still get it right. It's like saying, well when it is lined up perfectly when it is in a template it misses slightly when it isn't in a template. But you have the broad side of a barn to shoot at! How can you miss? Just move it down a bit. 199.125.109.64 (talk) 23:56, 12 April 2008 (UTC)

But I still think that what caused the problem is that the space where the title goes got wider. 199.125.109.64 (talk) 23:59, 12 April 2008 (UTC)

That wouldn't work when the sitenotice is active. You see, there might be a whole barn to shoot at, but as long as superman is randomly pushing the barn around on your farmland, you can still miss quite easily. This is the only way to guarantee where the element is drawn all the time, even when scaled and even when there is a (hypothetical) page long sitenotice. --TheDJ (talk • contribs) 00:16, 13 April 2008 (UTC)

Good point. One thing I will say it that it is better below the line than it was above the line - which looked all fine and good unless you had a really long place name, which happens a lot of the time, and it clobbers the coordinates. I would just like to see it tweaked just a little bit lower. Like 4.0 em or 4.5 em instead of 3.5 em. Also since the title space got bigger it could possibly even be located above the title - for some odd reason there is a new space above the title, and would mostly be above a long page title. Another thought, if sitenotice (whatever that is - is that the donation beg?) is active would that just mean that there are two locations that have to be aimed for, sort of like the side of the barn or the roof of the barn? 199.125.109.64 (talk) 01:03, 13 April 2008 (UTC)

I've been working on an extension that should make adding little icons up there more uniform and less buggy - hopefully we can convince the devs to install it when it's done. :) krimpet✽ 00:56, 13 April 2008 (UTC)

Here's another better idea to fix the problem of the coordinates getting covered up - just make the horizontal line only go half way across the page - problem solved. 199.125.109.64 (talk) 01:09, 13 April 2008 (UTC)

Whoever just moved the coordinates up a tad (to 3.0em?) moved them in the wrong direction. I just did a comparison of the space for the title between Wiktionary and Wikipedia, and Wiktionary has about the same code but has only a 34 pixel space while Wikipedia has a 45 pixel space, which like I say is I believe larger than it used to be. 199.125.109.131 (talk) 04:47, 16 April 2008 (UTC)

I don't know a lot about how the page is created, but I did find an old screenshot with the beg notice and the title, including the sitenotice, was about 134 pixels. I did notice a blank line in the code that had no effect, but which has now been removed (note the blank line between "content" and "top"):

<body class="mediawiki ns-4 ltr page-Wikipedia_Village_pump_technical">
	<div id="globalWrapper">
		<div id="column-content">
	<div id="content">

		<a name="top" id="top"></a>
		<div id="siteNotice">

The extra space is getting put in by the code that puts in the line. Taking out the above blank space has no effect. 199.125.109.88 (talk) 04:39, 18 April 2008 (UTC)

So has anyone looked at what changes were made to all the .css files in March to cause the extra space above the H1 title to appear? When I try to view the page locally none of the images in the background or the horizontal lines or the tabs show up, because I think they are all put in by all the style sheets rather than by straight html. In the meantime I see no reason for not compensating by just moving the coordinates down a space. I'm sure that someone will eventually find the edit that put it there. 199.125.109.80 (talk) 21:05, 21 April 2008 (UTC)

This isn't a fix per se, but the coordinates are included twice in the Miami, Florida article— once in {{Infobox Settlement}} and once in {{geolinks-US-cityscale}}. It is the latter that places the coordinates in the header. --— Gadget850 (Ed) ^talk - 21:20, 21 April 2008 (UTC)

Most geo-located articles add the coordinates to the header (title) only. There are many infoboxes that place the coordinates in both places and some articles have coordinates as a link at the bottom of the article that says something like Coordinates for XYZ are 00:00:00 N 00:00:00 E. The coordinate template {{coord}} template is normally located at the bottom of the article so that it can be found easily, but it is normally used only to place the coordinates up at the top right of the article along the title line. That's occasionally confusing to editors who can't figure out how the coordinates get there. 199.125.109.58 (talk) 21:39, 23 April 2008 (UTC)

Serving SVG images themselves instead of rasterizations

I use Firefox 3, and when I scale the images on a Wikipedia page, they become blurry and sometimes jagged. This happens even if the source image is SVG, because Wikipedia serves me a PNG rasterization instead. If Wikipedia served me the SVG file, Firefox could scale it much better, but I don't see any option to make Wikipedia do this. Will this be implemented in the near future? —Keenan Pepper 20:02, 14 April 2008 (UTC)

There are two main issues here.

1. It might be much more efficient (bandwith wise) to transfer a small png thumbnail than a detailed SVG image.
2. The client support for SVG is not "ubiquitous" enough yet to make this a default any time soon.

However, huge advances in SVG support (esp. Firefox and Safari) have been made over the past year. It is not entirely unlikely that we may see something this year where these clients will get full blown SVGs under certain conditions (although I doubt it will debut on en.wiki.x.io). More likely is however that the quality of the "thumbs" produced by the svg2png software will increase and that you are less bothered by the problem. --TheDJ (talk • contribs) 23:49, 14 April 2008 (UTC)

This is T5593. —Simetrical (talk • contribs) 16:12, 15 April 2008 (UTC)

I have created a userscript that loads SVGs in Safari 3. Works pretty well, but beware of the bandwith, because it will actually download the PNG and subsequently the SVG. Not really effecient. :D Still nice if you want to get an idea of course. I also tried to implement this for FF and Opera. Unfortunately they do not yet support the img tag for SVGs. I tried loading SVGs with object/embed/frame for these browsers. This works, but you get the size as defined within the SVG. I could not find a way to "resize" the SVG to our desired "thumb" size. In the past you could apparently do this with some Javascript, but all these methods seem to no longer work (probably in order to fix security issues). Anyways, for those who want to give it a spin on Safari 3 importScript( 'User:TheDJ/svg.js');. --TheDJ (talk • contribs) 23:34, 15 April 2008 (UTC)

Keep in mind that many SVG files on Wikipedia are *HUGE*, many times larger than the PNG rasterizations even if we did serve them compressed. Page loading and rendering is going to be a *lot* slower using inline SVGs. --brion (talk) 18:54, 16 April 2008 (UTC)

I too have noticed that Wikipedia resamples/scales PNG images much better than SVG images. I think there could be a simple fix to that: Perhaps let the MediaWiki software first render a big PNG image from the SVG, and then use the PNG rescaling function to make all the smaller images that actually get displayed. Note: I mean the MediaWiki software should do this, not that we should do this manually. Although manually uploading a big PNG version of a SVG for some icons that look bad can be a workaround.

--David Göthberg (talk) 13:24, 19 April 2008 (UTC)

Well, that certainly ought not to happen; SVGs should generally render fairly nicely at any size (though that doesn't always mean they'll look ideal at a size they're not designed for). Can you give some examples? --brion (talk) 17:26, 23 April 2008 (UTC)

As much as we might not like it we still have to use PNG over SVG to support IE (that rhymes!), which holds a good sized chunk of the browser market. User A1 (talk) 15:56, 23 April 2008 (UTC)

If we did start serving SVGs directly, we would of course use fallbacks to PNG support for compatibility. Unfortunately a lot of browsers seem to do this wrong (at least last I tested it, a couple years ago) and you tend to get prompts for plugin installation or otherwise general breakage for those which don't support SVG natively. As long as it wasn't an ideal situation (due to breakage, ugly messages, or too-large files), it would probably be relegated to a non-default user option if included at all. Ah, life on the web. :P :) --brion (talk) 17:26, 23 April 2008 (UTC)

Red diffs

Some (but not all) diffs are shewing with the new text in red - what is this please? DuncanHill (talk) 15:21, 18 April 2008 (UTC)

Could you give an example and say what behavior you expect? Diffs have always shown certain types of added text and removed text in bold red so that it stands out. —Simetrical (talk • contribs) 15:33, 18 April 2008 (UTC)

I believe he is talking about the dotted borders sorrunding the changed text, as in this diff. I don't remember seeing this before, but seems a handy feature. Especially for edits where only spaces, tabs and line breaks are edited. --soum ^talk 15:43, 18 April 2008 (UTC)

I mean like this [8]. DuncanHill (talk) 15:47, 18 April 2008 (UTC)

I don't recall ever seeing any red in diffs before (btw, I use the greenscreen gadget). DuncanHill (talk) 15:50, 18 April 2008 (UTC)

And the behaviour is the same with the gadget switched off, a few diffs ahew red text, most don't. DuncanHill (talk) 15:53, 18 April 2008 (UTC)

Also, there is a new class .diffchange-inline, which broke my existing CSS overrides. --Random832 (contribs) 15:50, 18 April 2008 (UTC)

I find these new additions useful and welcome them. Gary King (talk) 20:15, 18 April 2008 (UTC)

Red bolded text has always been there. The dotted line however... Blech! I've always used a custom red background for changed text. Try this in your monobook.css:

.diffchange {
  background: #fdd;
}
.diffchange-inline {
  border: none;
}

— Edokter • Talk • 21:43, 18 April 2008 (UTC)

I now see the reason why the border was added (which works just as well with the background); (leading) spaces are now visible in the diffs! — Edokter • Talk • 22:18, 18 April 2008 (UTC)

I'm torn. Seeing spaces is nice, but I find the border obscuring and distracting... which for me is a bigger issue than not being able to see the spaces, I'm afraid. Interesting thing to try, let's see how it shakes out and if people like it. – Luna Santin (talk) 01:15, 20 April 2008 (UTC)

Thanks for the script, works great. I admit that seeing spaces is nice indeed, but to read a big chunk of text all bordered is too annoying. Garion96 (talk) 01:20, 20 April 2008 (UTC)

Would be more difficult to run, but perhaps if we applied a second class when the number of characters changed is very small (say, one or two?), and that second class could add the border? Would make it easier to spot spaces, commas, and such, hopefully without interfering with our ability to read the rest of a diff. – Luna Santin (talk) 01:36, 20 April 2008 (UTC)

Perhaps another idea might be worth mentioning, even if only as an idea... To concentrate some resources on making wikEd usable by browsers other than Mozilla; its "Changes" preview tool is exceptionally good, in my opinion, and I think others will agree with me. Waltham, The Duke of 08:01, 20 April 2008 (UTC)

If possible that would be great. Garion96 (talk) 19:38, 20 April 2008 (UTC)

• There are several changes: dotted border, inconsistent red font on simply added lines (sometimes it's red, sometimes it's normal black like it was before), and it seems like MediaWiki now makes comparison across several lines (look at user signature in this diff. —AlexSm 22:40, 18 April 2008 (UTC)

Aaron has reverted this in r33580. —Simetrical (talk • contribs) 02:28, 22 April 2008 (UTC)

I talked to some people on IRC - this (multi-line comparison etc) is a result of changes in the PHP diff engine (in DifferenceEngine.php) which shouldn't even be being called on wm for performance reasons. Presumably it will be gone once they fix whatever's causing it to fall back on the PHP code. --Random832 (contribs) 02:54, 24 April 2008 (UTC)

Dotted borders

The dotted borders in diffs are not a good addition; they create confusion. Please remove them. Badagnani (talk) 23:13, 18 April 2008 (UTC)

Apparently it was one of the developer's ideas. I overrode the rule and the diffs should be back to normal now. —Remember the dot ^(talk) 01:16, 19 April 2008 (UTC)

I don't know; I like them, personally, because it is easy to spot small changes like an added or removed punctuations. Gary King (talk) 01:22, 19 April 2008 (UTC)

The distracting dotted red border still shows up. Please remove it, as it creates confusion and is very distracting. Badagnani (talk) 03:44, 19 April 2008 (UTC)

Have you tried bypassing your cache? —Remember the dot ^(talk) 04:14, 19 April 2008 (UTC)

I'm very familiar with this argument/instruction, which is brought out whenever a less-than-optimal change is made by the tech people. However, it shouldn't be necessary for me or our millions of other users, for all of whom this feature is distracting, confusing, and unnecessary. Please remove it, thanks. Badagnani (talk) 04:18, 19 April 2008 (UTC)

Sysops such as Remember the dot are not capable of overriding your cache, however much that annoys you. There's no point in complaining to them about it. When developers change the software's styles, they normally override everyone's caches by providing a new file name, but sysops don't have that option. —Simetrical (talk • contribs) 02:36, 22 April 2008 (UTC)

The dotted borders are, in fact, extremely useful for people who are even partially colour blind. —Ian Spackman (talk) 04:47, 19 April 2008 (UTC)

• Remove - Dotted borders, which are distracting and cause confusion. Let such highly specialized users set this up specially on their computer, the way any special-needs users might do. Badagnani (talk) 04:55, 19 April 2008 (UTC)
• Some of these diffs[9] are just awful. I reverted the CSS change from SVN. Maybe the developer who added it can come up with some compromise. Anyway, Remember the dot's edit should have it locally fixed for now. Voice-of-All 05:56, 19 April 2008 (UTC)

It's not been fixed. Please fix it. Badagnani (talk) 19:49, 19 April 2008 (UTC)

The site does not run directly off SVN. A sysadmin needs to synchronize the site to the SVN repo for it to take affect. Voice-of-All 19:54, 19 April 2008 (UTC)

Bah, someone reverted Remember the dot's edit. Well, either way, it should return to normal next SVN up/sync. Voice-of-All 20:00, 19 April 2008 (UTC)

Time out!

I have reverted the change that overrode the new border; common.css is not the place to store your personal preferences; you all have your onw monobook.css for that. Now... if there is to be a poll to gauge support for this change, people will have to be able to see what we're talking about; that is why I reverted.

However, if you are truly distracted by the change, all one has to do is place the following code into their own monobook.css:

.diffchange-inline {
  border: none;
}

If you want a lovely red background, you can also add this:

.diffchange {
  background: #fdd;
}

But PLEASE, don't edit common.css in a knee-jerk reaction to any change that you don't happen to like. — Edokter • Talk • 09:20, 19 April 2008 (UTC)

This feature has alredy been reverted in SVN - what's the point in your revert when it has effect for a couple days maximum? MaxSem^{(Han shot first!)} 11:57, 19 April 2008 (UTC)

I only noticed that later. In that case, removing unnecessary code is never pointless. — Edokter • Talk • 12:07, 19 April 2008 (UTC)

I still see this feature in some diffs, but not in others. Seems it only works partially. Anyway, I like this addition since I need to see small things like an added dot or a removed space when I check changes people do to templates. Yes, a removed space sometimes breaks a template. But I experimented with it to make it less intrusive. To me just underlining the change looked better. A red line or a dark yellow-green line worked best for me. Since the diffs show yellow background to the left and green to the right, thus yellow-green is in-between and looks soft on both backgrounds.

Red underlining:

.diffchange {
    border: none;
    border-bottom: 1px solid #f00; 
}

Yellow-green underlining:

.diffchange {
    border: none;
    border-bottom: 1px solid #6AD500;
}

If you want to try it, paste one of those codes into your own monobook.css. Then you need to refresh your web browser cache and wait 30 seconds or so before you load a diff to take a look. (I don't know why the wait is necessary, but it perhaps has something to do with Wikipedia page caching or database propagation.)

--David Göthberg (talk) 14:41, 19 April 2008 (UTC)

Your underlining system is good and deserves further thought. But before we jump into it, I believe I've found a bug in the MediaWiki software that needs to be fixed. Inserting whitespace is caught and displayed with class="diffchange diffchange-inline", but deleting whitespace is not. Furthermore, the diffchange class uses "white-space: -moz-pre-wrap" instead of the more universal "white-space: -pre-wrap". —Remember the dot ^(talk) 01:33, 20 April 2008 (UTC)

I've filed bug 13797 about the pre-wrap problem. It should be pretty trivial to fix, provided that we find a developer who's willing to fix the problem. —Remember the dot ^(talk) 01:42, 20 April 2008 (UTC)

The "old" way (without red dotted border) was just fine and not distracting; please restore that, thanks. Badagnani (talk) 08:06, 20 April 2008 (UTC)

I thought the dotted box was wonderful! Would I be right in thinking I can get it back by putting the diffchange in my monobook.css with nothing between the curly brackets? (I'm asking because I just tried it, but nothing happened yet; maybe I haven't waited long enough.)

I have a proposal: I think this is such a great idea, for those who want it, that it should be an option in "my preferences" under the "edit" tab. In order to make its purpose clear, there should be an illustration showing what it looks like, vs. the default. Maybe several options, including the underlining and coloured background ideas suggested above, could also be included, and examples shown of each. Where would be the best place to make this an official suggestion? --A Knight Who Says Ni (talk) 13:32, 20 April 2008 (UTC)

Why don't you make this available as a gadget? – Nikerabbit (talk) 14:12, 20 April 2008 (UTC)

Answer to part of my own question, FYI: yes, putting "diffchange" in monobook.css with nothing between the curly brackets brings back the boxes. Yay! --A Knight Who Says Ni (talk) 14:24, 20 April 2008 (UTC)

Incidentally, I've filed bug 13804 about the whitespace symmetry issue. —Remember the dot ^(talk) 23:23, 20 April 2008 (UTC)

Aw, now it's stopped working again. I'll bet the programming team completely removed the feature, which was not necessary, since it was over-ridden with the monobook parm. (Unless they removed it to resolve the whitespace issue mentioned above.) Hoping they will consider re-instating it as a User Preferences option, but that's a bigger change, so I don't expect to see it soon, if ever. But I did make a Bugzilla request for it. --A Knight Who Says Ni (talk) 12:58, 21 April 2008 (UTC)

Aaron removed it in r33580. —Simetrical (talk • contribs) 02:36, 22 April 2008 (UTC)

My own monobook.css uses

.diffchange {color:black; background-color:#FF7458; border:none;}
.diffchange-inline {color:black; background-color:#FF7458; border:none;}

and this seems to work well, though it's my preference. Just an idea. :) Nihiltres^{t.l} 15:04, 22 April 2008 (UTC)

Nikerabbit: I did add the underlining and the dotted border as preferences gadgets today. And it worked perfectly. But soon after I got reverted by another admin who seems to think he owns the gadget system. Read more about it below at Gadget discussion, and anyone is of course welcome to follow those links and join in the discussion there.

--David Göthberg (talk) 18:03, 23 April 2008 (UTC)

Job queue?

Hello, I am just curious why job queue is something like 8 million (en WP has only 12 million total pages). About a week ago it was like 6 million. What's going on? Renata (talk) 03:26, 20 April 2008 (UTC)

Where do you see these numbers? I do not see them on the page. Gary King (talk) 06:43, 20 April 2008 (UTC)

They're at Special:Statistics, and the job queue is now nearly 9.7 million. I have no idea why it is so high. Graham87 07:12, 20 April 2008 (UTC)

Well, one reason might be that the last few days we have changed the code of a lot of templates. Templates that in turn are used on LOTS of pages. Just one such example is that we are working through this list Wikipedia:Template documentation/List to add /doc subpages to many templates. You can see some other big jobs mentioned here: Wikipedia:Bot requests.

And some day from now we are going to update the {{ambox}}. (It is used on 343,000 pages...) The last few times we did that some servers ran out of memory and crashed. Although some claim that was just a coincidence. Let's see what happens the next time. And soon the {{navbox}} and {{navbox/core}} are going to be updated. Those are used on 615,000 pages. So hold your hats!

--David Göthberg (talk) 10:40, 20 April 2008 (UTC)

But isn't that /doc subpages are added in between nowiki tags and that should not affect the job queue? Ambox is popular, but still nowhere near 9 million pages... Renata (talk) 14:32, 20 April 2008 (UTC)

Does this activity impact on the speed of ?action=purge's, image updates and various other miscellanea? Should be be expecting that some things won't be updating so quickly over the next few days? Franamax (talk) 10:45, 20 April 2008 (UTC)

No, the job queue runs in the background, it shouldn't slow anything down appreciably. Prodego ^talk 13:59, 20 April 2008 (UTC)

I'm thinking in particular of things that go onto the job queue and will be placed inline with the other nine million tasks. Isn't that where (for instance) image thumbnails get updated? Or to ask a different way, what exactly does go to the job queue? Franamax (talk) 14:07, 20 April 2008 (UTC)

Yes, sort of. Purging will update everything on a page. Normally, if something like an image or template on that page changes, it is added to the job queue, and once it reaches number one on the queue that image/template is updated on that page. Prodego ^talk 01:07, 21 April 2008 (UTC)

The job queue is mostly used to run updates of link records due to template changes. That is, if your template changes to include more links or another template, then the job records go and update that so the link records are correct, and so that updates to that other template update pages correctly. --brion (talk) 01:24, 22 April 2008 (UTC)

I tracked down the problem with the extra-growing queue... slow clock on the database master caused false lag readings, which caused all the job queue batch processes to freeze waiting for the slaves to catch up. Fixed up NTP, resynced the clock, and it's purring along again. --brion (talk) 00:28, 23 April 2008 (UTC)

Ah... Am I correct in asuming that that's also what was causing the very frequent database lockups during edits in the last week or so? — Edokter • Talk • 00:37, 23 April 2008 (UTC)

Yes I expect so, and also why MelonBot spent three hours last night trying to get an edit in at maxlag=5 (I had to manually override it to maxlag=20 to get it to register!): human edits get a much higher maxlag limit IIRC, but if the clock issue was erroneously adding five or six seconds to every reading, then it would hit the lockdown limit much more easily. I watched the maxlag values for about five minutes last night, and it never dropped below seven seconds, so I guess the clock was about five or six seconds slow? Happy‑melon 11:04, 23 April 2008 (UTC)

Exactly; you'd see a minimum 7-second lag all the time, as the master's clock was 7 seconds slow. --brion (talk) 17:28, 23 April 2008 (UTC)

New Diffs Software ?

Is there a new version of the diffs software in use? I ask because I just went to WikiMedia's Bugzilla and reported a problem with the diffs report (see my comments in Bugzilla:10511) then found a similar bug -- with a better description -- under Bugzilla:9533. A little more searching found Bugzilla:9112 which really concerns me since it indicates that it may have been the WikiMedia software that corrupted the edit.

Anyone else seeing anything like this? -- Low Sea (talk) 11:46, 21 April 2008 (UTC)

The diff software in use has been stable for a couple of years now. It's always had occasional hiccups. —Simetrical (talk • contribs) 03:01, 22 April 2008 (UTC)

The term stable[10] implies the softare has not been changed because every programmer knows that even the smallest of changes can unintentionally de-stabalize the best software (those "occassional hiccups" represent we are not dealing with "best" yet). A few sections above this one there are a couple of discussions[11][12] about new and unexpected "red dotted boxes" in diffs which clearly indicates that the diff software is currently undergoing changes. -- Low Sea (talk) 13:29, 22 April 2008 (UTC)

The diff backend on Wikipedia is wikidiff2, and you can see for yourself how stable it is in its changelog. The last change that was more than editing the value of constant strings or tweaking makefiles seems to be in May 2007, and that was still a matter of a couple dozen lines, still adjusting details of the output only. The overwhelming majority of the code has not changed in over two years, since it was first written.

So no, the adjustment of string literals in two places in the code to output extra CSS classes is not going to "de-stabilize" the software in any way that could explain the bugs you mention. Especially since the changes made were a few days ago, whereas the bugs you report are months old; and for all I know the software wasn't even recompiled on Wikimedia.

What exactly causes the bugs I don't know. They aren't consistently reproducible; refreshing the diff usually causes them to disappear. As every programmer knows, such bugs sometimes occur in even the best software and can be extremely difficult to track down. Since they're quite rare, it's not necessarily worth the effort. T11112 doesn't seem to be a problem with the diff software at all, but with the stored revisions themselves, and may be due to user error (as Brion suggests there). —Simetrical (talk • contribs) 15:20, 22 April 2008 (UTC)

I was not familiar with where to look for the software change logs, thank you. You make a valid point. I would generally tend to agree that "mere" string changes ought not impact software (although it can if the parser is buggy -- an excellent example of that is the MSIE parser which caused infamous buffer overflow errors depending on the URL string) and I certainly agree that the timeframe does eliminate the April change as suspect for the root cause (though it might have aggravated the cause if it is parser based).

The January revision 30294 looks pretty extensive, and specifically appears to impact diff display protocols related to EOL. This may or may not be part of the problem. I know it will take time to find the bug especially if it is intermittent. One thing that might help here is that the diffs I was reporting are still consistently erroneously under-reporting the actual change. If you look at this diff[13] then look at the full page before[14] and after[15] you can see the diff reporting software is consistently under-reporting the changes. I might suggest a monitored step-thru of the wikidiff2 code using these two pages as data to see why it is suppressing diff lines. -- Low Sea (talk) 21:41, 22 April 2008 (UTC)

First of all, r30294 changed only one file in wikidiff2, and that only changed metadata (see actual diff). It changed the stored formats of various files so that they used SVN's "eol-native" flag to automatically provide Windows newlines to Windows users and Unix newlines to Unix users, according to the convention used for all software in our repository.

As for the diff you link to, in fact, it appears to be entirely correct. When I run diff -u on the wikitext source of each revision, copy-pasted to my home computer, I get identical results to the diff that's currently displayed:

--- /tmp/diff1  2008-04-23 12:56:06.000000000 -0400
+++ /tmp/diff2  2008-04-23 12:56:10.000000000 -0400
@@ -67,18 +67,7 @@
 
 === Tenets ===
 Julie Ann Storr, founder of Nibbana ([[Sydney]]) in a how-to of the film's tenets, reports, "it all starts with gratitude"<ref name=nibbana_manifestation>{{cite news
-  | last = Storr
-  | first = Julie Ann
-  | title = Open the Gates of Manifestation
-  | language = English
-  | year = 2006
-  | month = 11
-  | url = http://www.nibbana.com.au/page.aspx?id=64
-  | accessdate = 2007-06-10
-}}</ref> and Stephanie Whittaker of [[Montréal]]'s ''[[The Gazette (Montreal)|The Gazette]]'' notes, "proponents ... talk about a universal intelligence that responds to our desires.<ref name=the_gazette_montreal /> The film encourages the viewer to see "the Universe [as] 'a catalog' that we can flip through and shop"<ref name=beliefnet /> and advises surrounding oneself with "positive" people.<ref name=tmcnet /> [[Visualization]] and [[Vision boards]]—anything on which one has placed images of what one wants—are recommended as aids for manifesting desires.<ref name=Calgary /> Paul Harrington, the co-producer, uses his computer's screen saver as a vision board.<ref name=bleeping_herald /> ''The Secret'' lists three required steps — "ask, believe, receive" — as the essence of the Law of Attraction:
-
-{| class="wikitable" width="80%"
-!| Step<sup>†</sup>
+  | l
 !| Commentary by Gazette<ref name=the_gazette_montreal />
 !| Commentary by Nibbana<ref name=nibbana_manifestation />
 |-

Possibly when you viewed the diff it was incorrect. As I said, these issues are not easily reproducible. If they were, we could in fact just conduct "a monitored step-thru of the wikidiff2 code". We know that. Give us some credit. —Simetrical (talk • contribs) 17:03, 23 April 2008 (UTC)

OK, I think I have found the "problem" and it is not with wikidiff2 but rather with the REF tag parser. It is pretty ugly and I do not know if this is fixable behaviour.

When I visually look at the pages I see enormous differences yet when I look at the diffs I see very little. On the strong assumption that you were very correct about "when you looked at the diff it was incorrect" I attempted a couple experiments:

First, I created a sandbox page with the old wikitext, then replaced it with the new wikitext, then looked at the diffs. They showed exactly the same as on the original article page diffs. Since this is a new edit and a new diff I feel safe to assume that the diff software is behaving identically both now and back when the original problem occured. While I had actually been hoping you were right, this pretty much eliminates the idea that the diff was wrong for any given point in time.

Second, I compared the raw wikitext on my home machine using CMP32. Now my diff software is a not the same and I was not expecting identical results but the report was close enough to wikidiff2 for this test. Apparently wikidiff2 and cmp32 produce substantially similar results.

This left me with a problem. My eyes were still showing me something was wrong. So I started looking at what was deleted and eventually it hit me that the problematic edit had deleted part of the <ref> citation including the closing </ref> part. So Wikimedia simply is treating everything up to the next </ref> as part of the citation and suppressing significant amounts of data because the ref parser is not complaining about garbage data. Can this be fixed ? Can the badly formatted ref tags somehow flag the user when saving/previewing ? -- Low Sea (talk) 20:42, 23 April 2008 (UTC)

The current behavior of refs is not so straightforward to fix. If the closing </ref> is missing, then how is it supposed to know where to stop? The literal string "<ref>" could legitimately occur in a reference's text. Not really a bug, unless you have a solution to propose. —Simetrical (talk • contribs) 00:27, 24 April 2008 (UTC)

I was afraid of that when I said this may not be fixable. The only minor solution I might recommend is to determine a reasonable byte count for data between the <ref> and the </ref> and then have some kind of "are you sure" pop up when saving/previewing if the number seems excessive. Also, if the ref is using a WP citation template then again have the editting software question if the template contains unknown keywords. -- Low Sea (talk) 01:04, 24 April 2008 (UTC)

Also, could the diff software be made to show when text goes from visible to hidden as a diff? I realize that is probably pretty complicated. -- Low Sea (talk) 20:48, 23 April 2008 (UTC)

Diffs are wikitext source-level only. There are no plans to make them show HTML diffs at present. —Simetrical (talk • contribs) 00:27, 24 April 2008 (UTC)

A different DIFF problem

There seems to be another problem. Look at this diff. Several lines are replaced with one single line, but the diff is showing them to be replaced with blank lines. Is this a known bug? --Jean Lugan (talk) 16:09, 23 April 2008 (UTC)

This is not a bug; it's completely correct. The first line is listed as being replaced by the first line of the replacement, and the rest are shown as deleted. That's just how the side-by-side diff works: it associates a single line on the left with a single line on the right. If there are more lines on the right than on the left, it shows the leftovers as deleted, and vice versa. —Simetrical (talk • contribs) 17:03, 23 April 2008 (UTC)

What I see when clicking on the diff are the same number of deleted (left with yellow background) and added (right with green background) lines. To be correct it should look more like this one: The deleted lines are represented by white background on the right. Can you see the difference between those diffs or is this a problem with my browser? --Jean Lugan (talk) 17:16, 23 April 2008 (UTC)

I see the right-hand lines with white backgrounds in both links (other than the actual added lines). —Simetrical (talk • contribs) 00:27, 24 April 2008 (UTC)

Category:Non-free images lacking article backlink

This category is full of stranged use of the template like {{Hpfur}}. I also think some of userspace versions are used for transclutions. I started to clean them up when I thought they were all just archives and examples, but now that I think some are being used I am afraid of messing things up. Will someone more technical straighten this out?--BirgitteSB 21:23, 22 April 2008 (UTC)

Well, since it bothered you so, I fixed all the ones in my userspace that found there way into that category (and you are quite right that mine were being used, my userscript uses them as template template trasnclusions, if that makes sense). If you are just worried about the other userspace examples, you might think about dropping a note on their talk pages, as I am sure some have just been forgotten and they would be fine with removing them. - AWeenieMan (talk) 22:02, 22 April 2008 (UTC)

Ooh er: {{hpfur}} is one of mine! I've fixed its appearance on the template page, although to be honest I don't think anyone would miss it if it were deleted. Perhaps a more thorough solution would be to set {{non-free use rationale}} to only categorise in the image namespace? Happy‑melon 11:00, 23 April 2008 (UTC)

If anyone knows how to set it up to only categorize image space that would be great. Otherwise I will fix the one's in archives by hand and leave notes about the others.--BirgitteSB 14:54, 23 April 2008 (UTC)

You probably just want {{#ifeq:{{NAMESPACE}}|Image|[[Category:...]]}}. —Simetrical (talk • contribs) 17:06, 23 April 2008 (UTC)

I implemented that on the master template, so hopefully these odd inclusions will (eventually) work their way out of the category. Happy‑melon 20:08, 23 April 2008 (UTC)

Templates: concatenation of strings

I'm trying to code a template which creates a paragraph with references, but the references aren't formatting properly. I need to be able to concatenate strings. The draft is at User:Gadfium/census template.

The problem lies in <ref name=2006Census>{{cite web|url=http://www.stats.govt.nz/census/2006-census-data/final-counts/{{{3}}}.htm|title=Final counts – census night and census usually resident populations, and occupied dwellings - {{{4}}}|publisher=[[Statistics New Zealand]]}}</ref> where the parameters aren't being substituted. I tried using {{#vardefine:variablename|value}} which didn't seem to help, and I thought {{param}} might be a suitable way to concatenate strings, but no joy.

The template is not so much for use on the English Wiki but on the Maori one. I don't speak the language, but I have translations for the specific phrases used, and a working template will reduce the likelihood of errors adding populations to localities. For an example of the desired final result, see the second paragraph of Ngunguru.-gadfium 22:13, 22 April 2008 (UTC)

See the last in WP:FN#Known_bugs. I think what you're trying to do won't work with the order MediaWiki currently parses pages. You can use inline links for the references rather than using ref tags to avoid this issue - this works reasonable well as part of a table, for instance. Gimmetrow 22:20, 22 April 2008 (UTC)

Thanks. I'll just add the references separately from the paragraph.-gadfium 22:57, 22 April 2008 (UTC)

Gah, there is a workaround. Use the #tag magic word for the ref tag. Nihiltres^{t.l} 23:07, 22 April 2008 (UTC)

Thanks very much.-gadfium 23:51, 22 April 2008 (UTC)

Oh, that's relatively new. The last bugzilla:2257 entry from 26 March 2008 says this bug is still open. What's the standing issue? Gimmetrow 01:13, 23 April 2008 (UTC)

It should Just Work(TM), not force people to write everything using curly braces. —Simetrical (talk • contribs) 17:09, 23 April 2008 (UTC)

Change Management

Just out of curiosity is there a Change Management system in place for the WP software? As complex and as seriously impacting as changes can be one would expect such a system even in an open source environment. (By the way, the WP article on Change Management needs serious improvement in both style and substance.) -- Low Sea (talk) 01:01, 23 April 2008 (UTC)

The MediaWiki software works by means of sub-versions within a well-defined change management system. It looks a little funny just at the moment (which I need to check out), but generally works well. See Special:Version for what is running at the moment. Franamax (talk) 01:30, 23 April 2008 (UTC)

Oops, I asked the wrong question. My bad. What I meant to ask was:

Is there a Change Control process here?

As the WP page on Change Control says "There is considerable overlap and confusion between change management, configuration management and change control.", however the process described on that page does correctly describe the concept I am asking about. -- Low Sea (talk) 05:58, 23 April 2008 (UTC)

There are relatively few developers; behind the scenes we're a small operation, although we service one of the largest websites there is. Therefore, we don't have overly formal procedures for most things. Generally speaking, developers (those with commit access) commit anything they want to the Subversion repository, without any review process. (Of course, committing things that don't work is frowned upon, and those are backed out.) The lead developer, Brion Vibber, reviews all changes to the software as they occur, time permitting. Periodically, once a day to once a week, he (or, in theory, any of a number of other people) updates the software running on the servers to the latest version. Since it's only a couple of days' worth of changes, in most cases any problems that people report can be quickly tracked down and fixed.

You'd think it would be a little more formal than that, with a testing procedure and fancy stuff like that, but it's not.  :) Wikipedians get to be our guinea pigs. It's very rare that the entire site goes down (changes are briefly tested out on test.wiki.x.io before going live), but occasionally there are semi-significant mishaps, like certain pages failing to display just a little while ago. But they can be quickly fixed once reported, so it's not a big issue. We're a website, not a bank, and it frankly doesn't matter so much if the occasional edit goes wrong, purism aside. —Simetrical (talk • contribs) 17:21, 23 April 2008 (UTC)

Good answer! Not very comforting to an old availability manager like me (bet that is a red link hehe) but very nicely said! Thanks for the info. -- Low Sea (talk) 21:46, 23 April 2008 (UTC)

Complicated category transclusion not purging

I have created a fairly complicated template that makes use of parser functions to add a category (see commons:Template:Birthcat). The category seems to be transcluding as it should: see commons:Category:1761 births; it links to commons:Category:18th century births at the bottom. However, if you click on the 18th century births, it doesn't include the new categories, even after using the action=purge function. Help! 128.118.226.88 (talk) 07:56, 23 April 2008 (UTC)

A null edit to commons:Category:1761 births fixes this. Algebraist 08:03, 23 April 2008 (UTC)

Algebraist: No, that should make no difference. Edits, null edits and purging does not have an effect on the updating of the category listings.

128.118.226.88: MediaWiki runs updates of category listings as a low priority job. Thus if the servers are very busy it can take some days before a page is added or removed from the listing in a category. But usually it is updated within an hour. And I see that the category now have been updated.

We really need to inform people in some efficient way that category updates is run as a low priority job. I run into people all the time that are confused by it. Especially now when the servers have been busy and the delays have been more noticeable. Categories on user pages seems to be updated at even lower priority since they currently take over 4 days to update.

--David Göthberg (talk) 08:13, 23 April 2008 (UTC)

Not true, performing a null edit on any page forces a complete reconstruction and update of all relevant tables for that page, including the category links table. Happy‑melon 10:45, 23 April 2008 (UTC)

Well, you are almost right Happy-melon. Algebraist said a null edit (a purge) to "Category:1761 births", not the category it gets listed in "Category:18th century births". And that helps if the category the template is supposed to add is not shown at the footer of that category page. But 128.118.226.88 said it was showing, and then the only thing that helps is to wait. But purging "Category:18th century births" should not have any effect. (But just in case I tend to press my little home coded purge button, since you never know. And it feels better to at least have tried.:))

--David Göthberg (talk) 17:44, 23 April 2008 (UTC)

Thank you all. A null edit would not have helped, as it would have only updated that one category into 18th century births. However, all is good now, thanks for the response. The Evil Spartan (talk) 19:47, 23 April 2008 (UTC)

Thanks and question about logs of user input

Kudos and questions. Thanks, how fun--as others mentioned, something similar to the commons' mwsuggest working today on Wikipedia search. Live wiki-ing is on some people's wishlists, or so I have read. Apparently, at least in the commons, interaction is all collected down to details like mouse movement. At the present time, when users edit in an edit window, 1) are all those forward and back keystrokes, movements and changes recorded in the raw server logs, or, if that information is discarded 2) will it be recorded in the future? I'll guess no, that storage has gotten cheaper but not quite that cheap yet. Just wondering as I never did learn JavaScript but read about products that do this on commercial sites (I think). Thank you. -Susanlesch (talk) 10:14, 23 April 2008 (UTC)

The english wikipedia has 218 million edits in the database: if every cursor movement and character typed were to be recorded, the quantity of data to be stored would spiral out of control; as would the server time needed to process them. Besides, both typing and mouse movement is performed client-side and only recorded by the browser, it's certainly not sent to the server - I don't know if returning that data would even be possible. I think it highly unlikely that details like mouse movement is recorded on Commons, and such data would be monumentally difficult to analyse (as the locations of clickable links, buttons, etc, would be affected by browser type, version and settings, window size, personal stylesheets, and a host of other possibilities. Wildly impractical, IMO. Happy‑melon 10:42, 23 April 2008 (UTC)

Thanks, you're right. I guess commons and in most cases this is client side. At WWW2006 a group presented UsaProxy which adds the code need to capture anything JavaScript can describe with a few exceptions maybe. I assumed but really don't know if commercial sites do this when they need to. But for usability study they probably don't need almost 7,000,000 users. :-) -Susanlesch (talk) 11:04, 23 April 2008 (UTC)

No, we don't collect any data of that sort. Hypothetically we could do some such data collection for usability testing, but at this time we don't (and it would not be practical to do so in bulk, as mentioned above). --brion (talk) 17:20, 23 April 2008 (UTC)

Brion, thanks much for your reply and Happy-melon, too. I appreciate it. It's probably good to know that capturing details can conceivably happen; mwsuggest can do it locally now for quite some time at Wikimedia commons. In real life I can't imagine why anyone would want to save all that data to logs (the UsaProxy people for example chose to limit mouse coordinate recording to a sample per some number of milliseconds) but in some circumstances I guess an intruder could watch. Realistically though and for any practical and legitimate application, over the years Jakob Nielsen has taught "Why You Only Need to Test With 5 Users" and so on--that a surprisingly tiny sample of users is more than plenty for most usability studies. Thank you again and best wishes. -Susanlesch (talk) 21:30, 23 April 2008 (UTC)

Monobook.js

Resolved

The OP has blanked it themself. Algebraist 15:22, 23 April 2008 (UTC)

Hi, i added a javascript to my monobook.js - now i cant seem to log in. Not sure if it is the monobook that is the problem, but i cannot even revert the change. The browser just hangs and crashes, had to delete cookies to get onto this page without being logged in. Is someone able to delete my monobook.js ? Thanks User:Kennedygr —Preceding unsigned comment added by 194.81.124.196 (talk) 14:55, 23 April 2008 (UTC)

Thanks for taking a look anyway, no idea why that happened. Had to disable JS and try again after deleting cookies. Think it may be something to do with the script? Or with me? ← κεηηε∂γ ^{(talk) (secret)} 15:41, 23 April 2008 (UTC)

Performance and User:Pageview bot

User:Pageview bot is currently creating a user subpage for every article it is enabled for. This is already more than 10.000 pages (all pages in the WP:VG project) I know of, and probably more. Theoretically, this bot would have one user subpage for every article on Wikipedia.

Usually I am the last to worry about performance, but the numbers involved made alarm bells go off with me. Would it be a better idea to seek an alternative solution, or maybe even bin this idea and request a MediaWiki feature? User:Krator (t c) 15:28, 23 April 2008 (UTC)

Given that we have this site, I fail to see why that bot is necessary at all. Mr.Z-man 16:56, 23 April 2008 (UTC)

There is a feature (disabled here) in MediaWiki that tracks page views but it is disabled here due to caching issues. Nakon 17:01, 23 April 2008 (UTC)

It's crazy to create millions of user subpages for this. It would also cause errors if the user ever got renamed, because that just moves the pages one by one, and it would time out. Other than that I just have to ask what the point is. This is exactly what the toolserver is for. Wikipedia is not a relational database. Performance issues . . . yeah, maybe, if it did actually track all page views. But it seems like there are much better ways to do this regardless. —Simetrical (talk • contribs) 17:37, 23 April 2008 (UTC)

See open discussion at Wikipedia:Bots/Requests for approval/Pageview bot. Mr.Z-man 17:41, 23 April 2008 (UTC)

The bot has been blocked. Stifle (talk) 20:00, 23 April 2008 (UTC)

Sorting tools?

Are there any Wiki sorting programs or scripts that take surnames and pipes into account?

What about foreign characters?

I just got done expanding List of Iceland-related topics, but the people on the list are placed by first name. Sorting the list by hand will be a bitch (and will have to be done again as the list grows).

Terms with accented characters didn't sort in with the rest of the list, and I left those at the end.

Comments and suggestions welcome.

The Transhumanist 16:30, 23 April 2008 (UTC)

Why would it have to be sorted again when the list get larger? Just insert each item in the right place first time you insert them. We do that over at Wikipedia:WikiProject Cryptography/List of cryptography topics and that list is always fully sorted.

Anyway, another option is to separate persons from other things. Since only persons are sorted by the last word, right? Then you can sort the two lists separately in whatever sort program you have in your computer that can handle Icelandic characters. You can of course have the two lists on the same page, for instance humans in the upper part of the page and everything else in the lower part.

But then again, why make it so complicated? Over at the crypto topic list we simply sort by the article name, no matter what. So Ralph Merkle is sorted into "R", not "M".

And by the way, how do you mean a computer should be able to tell apart a human name from a name of something else, if the human name is an unusual one?

--David Göthberg (talk) 17:24, 23 April 2008 (UTC)

Mass querying the articles to see if they belong in any humanesque categories, perhaps... ? (not a serious suggestion, btw) It's usually possible to sort a list people's names by last name with some simple heuristics (one catch is checking for lowercase multiple-word surnames like "von Grimm"), but as David said, I don't believe one could realiably differentiate between humans and non-humans. Gracenotes^T § 18:11, 23 April 2008 (UTC)

Random updating of my own sig?

I just noticed that, when replying in a thread that I've already posted on, I'm suddenly seeing the &#9775; of my signature being converted to "☯". &#9775; is supposed to display as ☯, but I'm somehow converting the text (I've hit the edit button and gone straight to "Show changes", without making any changes, and it pops up there).

When I log out and try to edit the same section, I don't automatically convert anything. I haven't done anything, and haven't made any changes to my monobook. I'm confused. :\ EVula // talk // ☯ // 18:19, 23 April 2008 (UTC)

So... you first had "☯", and now you are seeing "&#9775;"? All shows up ok with me anyway. - Face 18:56, 23 April 2008 (UTC)

No, I'm apparently rendering all HTML entities whenever I post; I just noticed that I was rendering the ampersand in the &nbsp; of someone's sig, converting it to a space. Not good. :\ EVula's sock // talk // 19:40, 23 April 2008 (UTC)

Any custom Javascript or gadgets enabled, by chance? Something might be eating things. --brion (talk) 19:53, 23 April 2008 (UTC)

I cleared out stuff from my monobook in an attempt to find it, but hadn't gotten around to clearing the gadgets. After doing so, it appears that refTools is undeniably the culprit (deactivated everything and, one by one, turned them back on, including toggling refTools specifically to verify the findings). I'll alert Z-man of this thread. EVula // talk // ☯ // 20:39, 23 April 2008 (UTC)

Try using the version in User:Mr.Z-man.sock/monobook.js in your JS page. It might be related to the same bug that old versions of Firefox have, which that should solve (and will be integrated into in an update of the gadget shortly). Mr.Z-man 20:46, 23 April 2008 (UTC)

Restoring a removed image

A few days ago, I noticed that User:Maxim (who recently left WP) removed the image Pokemon-diamond-and-pearl-group.png, citing an invalid fair use rationale. I then asked Jéské Couriano if it is possible to restore it so that I have a chance to improve the FUR. He answered that he tried, but was unable to do so, which I find a bit weird. I was wondering if non-free images with copyright issues are permanently removed or something. Cheers, Face 18:49, 23 April 2008 (UTC)

Strange, I was able to restore it without any problems. EVula // talk // ☯ // 19:42, 23 April 2008 (UTC)

Thanks! I have added an appropriate rationale right away. If anybody here can further improve the wording, please do! Cheers, Face 22:17, 23 April 2008 (UTC)

User rights tab script

I am not sure which of the scripts I have makes the $ tab when viewing a user or user talk page (to check user rights) but it seems to be malfunctioning for me right now. No matter which user page or user talk page I'm on - I get " Example (Administrator)". I'm a bit puzzled by this. I notice that this edit added "*Example (Administrator)" to Special:ListUsers last week. Could that be the problem? --Doug.^{(talk • contribs)} 23:20, 23 April 2008 (UTC)

Broken refs detection

While experimenting with my home wiki I noticed that if you add a category to system messages related to <ref> citations, such as MediaWiki:Cite error ref no input, all pages with such errors will be added to that category. I suppose that such measure could help us to detect broken formatting. MaxSem^{(Han shot first!)} 13:34, 16 April 2008 (UTC)

I have all the ref errors documented at Wikipedia:WikiProject Scouting/References#Troubleshooting. --— Gadget850 (Ed) ^talk - 13:43, 16 April 2008 (UTC)

No I don't. Just took a look at my test page and I see that numeric errors are no longer used. Looks like an update is in order. --— Gadget850 (Ed) ^talk - 14:03, 16 April 2008 (UTC)

I think it's a good idea Max, just like we have a cat for pages that exceed the transclude limits, we should have a cat for this. And it's easy to do. --TheDJ (talk • contribs) 15:57, 16 April 2008 (UTC)

What would cause a Cite error ref no input? --— Gadget850 (Ed) ^talk - 16:51, 16 April 2008 (UTC)

• I wrote a template to be added to the messages at {{broken ref}}, please review before it goes live. MaxSem^{(Han shot first!)} 17:09, 16 April 2008 (UTC)

You need to list this at Wikipedia:WikiProject Inline Templates#New inline templates. You should look at {{dead link}} and see how it uses {{fix}} as a standard method for in-line templates. --— Gadget850 (Ed) ^talk - 17:34, 16 April 2008 (UTC)

It's not an inline template. It's not intended to be inserted in articles. It's for MediaWiki: namespace only. And it shouldn't replace existing messages, only tag pages that use them. MaxSem^{(Han shot first!)} 17:37, 16 April 2008 (UTC)

Then I don't understand how this is used. --— Gadget850 (Ed) ^talk - 18:15, 16 April 2008 (UTC)

It's to be inserted in MediaWiki:Cite error ref no input and so on. MaxSem^{(Han shot first!)} 18:19, 16 April 2008 (UTC)

Ok. Where does MediaWiki:Cite error ref no input come into play? --— Gadget850 (Ed) ^talk - 18:25, 16 April 2008 (UTC)

Ah- I see it now. Cite.php has an internal message "Cite error ref no input"; this is passed to MediaWiki:Cite error ref no input that then shows "Invalid <ref> tag; refs with no name must have content" in-line with the ref. If you add a category to the message page, then all error will be categorized. --— Gadget850 (Ed) ^talk - 18:35, 16 April 2008 (UTC)

• So— now that I understand how it works, I support it. The last time I dug into cite.php, it was using numeric errors and it took me a bit to realize it now remaps the error messages. --— Gadget850 (Ed) ^talk - 10:28, 17 April 2008 (UTC)

 Done, Category:Pages with incorrect ref formatting went live. MaxSem^{(Han shot first!)} 13:35, 17 April 2008 (UTC)

Pages are missing from the cat: User:Gadget850/Cite errors should definitely show, as should Jim Prentice, Declan O'Sullivan and more. --— Gadget850 (Ed) ^talk - 13:48, 17 April 2008 (UTC)

I think it is taking a while to trawl through all the articles. More keep turning up as I am clearing it whilst the articles haven't been edited in days. It will take a while to populate that is all. Woody (talk) 13:51, 17 April 2008 (UTC)

"The job queue length is currently 8,620,858. which might explain it! Woody (talk) 13:53, 17 April 2008 (UTC)

Ah- good point. I expect there are a lot of pages with these errors. I had documented cite errors on a project page at Wikipedia:WikiProject Scouting/References#Troubleshooting. Perhaps this should be moved to Wikipedia:Footnotes? --— Gadget850 (Ed) ^talk - 13:57, 17 April 2008 (UTC)

I just did a null edit on User:Gadget850/Cite errors and it caused it to turn up in the category. Purging was not enough. It can change the purged page but not other pages like categories. PrimeHunter (talk) 13:58, 17 April 2008 (UTC)

Yeah the job queue is running really slow right now - I'm waiting on Category:Featured articles, which I edited six days ago, but it still hasn't fully populated. Null edits on the whole mainspace? I think not :D Happy‑melon 21:22, 17 April 2008 (UTC)

Changes to MediaWiki messages will not trigger re-renderings of any pages that may include them, via the job queue or otherwise. Such messages' use is not tracked in the database in any fashion, the way link/template/category inclusions are. Pages won't be added to the category until they're next edited, or otherwise purged for some reason (e.g. changing an included template). I'm not sure if the parser cache automatically expires old entries. Anyway, needless to say, don't null-edit a million pages with a bot or anything, be patient. —Simetrical (talk • contribs) 15:23, 18 April 2008 (UTC)

No- the 'pedia has been running slow enough these past few days. The category seems to be filling up slowly but surely— over 200 articles yesterday and now 470. The real problem is what to do about the broken refs. I have sample about 10% of the articles; of these all have "Cite error: Invalid <ref> tag; no text was provided for refs named $1" indicating that a named reference was called but never defined. I'm guessing most editors do not understand what this error means (perhaps that is a next step: clarifying the error messages). Should they simply be deleted and replaced with {{fact}}? --— Gadget850 (Ed) ^talk - 16:28, 18 April 2008 (UTC)

No, because in many cases (I'd guess at least 50%) the problem is that the ref tag was originally working, but was broken when a different chunk of text (including the ref tag which actually contained the reference text) was removed. The reference can be easily fixed by finding the reference in the page history. Perhaps the error message should link to a help page explaining the problem and how to possibly fix it. We'd be shooting ourselves in the foot to consciously go from a "this reference is almost working, but is broken due to a probably-fixable technical error" to "this reference does not exist, please find one". Happy‑melon 21:02, 18 April 2008 (UTC)

The best thing to do if you see a broken ref like <ref name="where did this come from" /> is to comment it out (<!--<ref name="where did this come from" />-->) so it doesn't look bad, and hope someone finds the real reference (<ref name="where did this come from">It came from here</ref>). Of course, if it can't be found, {{cn}} will be necessary. Superm401 - Talk 00:30, 23 April 2008 (UTC)

The problem is, that takes it out of the category. I think we should be taking the time to actually fix these in all cases. I fixed one just now that was due to the reference being present in the source in a misspelled infobox template parameter. --Random832 (contribs) 16:35, 24 April 2008 (UTC)

Moving templates

If I moved a template to a new title (obviously leaving the automatic redirect in place), would it add all the pages the template is transcluded on to the job queue? Happy‑melon 21:20, 22 April 2008 (UTC)

Actually, it doesn't seem to do that. Voice-of-All 21:35, 22 April 2008 (UTC)

Why should it? —Simetrical (talk • contribs) 17:21, 24 April 2008 (UTC)

Main Skin Style for Foreign Languages

I just visited the Romanian language Wikipedia page and the main skin (a version of Monobook I guess) has rounded corners/headers/etc. I was wondering if this has been discussed before and anyone who thinks this looks better/is more functional than the current default skin for the English language page. First headers are easier to read (such as the main "navigation" and "interaction" headers), and to me it is just more aesthetically pleasing. Nucleargrass (talk) 21:39, 23 April 2008 (UTC)

Unfortunately, those rounded corners only show up with Firefox. Internet Explorer does not support them :-(. - Face 22:46, 23 April 2008 (UTC)

Doesn't show up in Safari, either. There are ways of making rounded corners, but I don't have the links handy, and I'm unsure of how easily implemented it would be on-wiki. EVula // talk // ☯ // 23:41, 23 April 2008 (UTC)

Well, maybe sometime in the future everyone will be able to enjoy those rounded corners, haha. Nucleargrass 08:58, 24 April 2008 (UTC) —Preceding unsigned comment added by Nucleargrass (talk • contribs)

Expand/collapsible watchlists

Also posted this on VP/P, but I thought you techies know more about this stuff :)
Don't you just hate it when you are watching a page but do not see the edit because another editor (or the same) made another edit to the same page therefore hiding the previous edit from your watchlist? Yes, there is a preference to expand your watchlist but that's just too much clutter. I'm proposing making watchlists like histcomb.js, where if there are more than one edit to a page, it displays a number next to it showing the amount of edits done since you've last visited the page. Clicking on it will reveal the edits. I've made an example to illustrate what would be ideal: -- penubag  (talk) 02:38, 24 April 2008 (UTC)

This would also help prevent against vandalism that slips between other edits.

"Expanded watchlist" option goes nicely with "Recent changes → Enhanced recent changes (JavaScript)" option, which collapses multiple edits, just like you're suggesting. I even think that any other combination of these two options makes watchlist unusable. P.S. Yes, this second option name is a bit misleading, just like the option next to it which says "Titles in recent changes" but in fact sets the number of titles on most special pages. —AlexSm 03:15, 24 April 2008 (UTC)

Do you mean a watchlist like this? What I really wanted is to be able to see all the edits done to a page instead of just the current version being displayed. Sort of like Preferences -> Watchlist -> Expand watchlist to show all applicable changes but only compact them into one watchlist entry. PS: histcomb.js is great!-- penubag  (talk) 03:40, 24 April 2008 (UTC)

(Thank you) Your first screenshot shows WL without "Enhanced recent changes" option, while on the second (linked) screenshot this option is on, but now "Expand watchlist" is disabled. What I'm saying, you need both options enabled: then you'll see all edits, and multiple edits made in the same day will be collapsed. The script user:js/watchlist appends (among other things) the ± link to expand them all, if you need that. —AlexSm 05:38, 24 April 2008 (UTC)

Oh, I see! Thank you Alex! Even though I prefer a version like my image above (for showing only the top contb), this will do. Thanks again -- penubag  (talk) 05:42, 24 April 2008 (UTC)

Wikia does this, and I simply love it. Make a watchlist over at any Wikia site to see what I mean. I asked once if it was an option for any MediaWiki installation, and I think (IIRC) that it was something that Wikia's techs did specifically. I'm sure if someone were to ask them nicely they would give us a hand in getting it to work for us (I might be thinking of another feature, though). -- Ned Scott 01:47, 25 April 2008 (UTC)

Here's a screen shot

-- Ned Scott 02:01, 25 April 2008 (UTC)

Your screenshot shows the same enhanced expanded watchlist that you can have here on Wikipedia (the one I was talking about above). Unless I missed something really special on that screenshot... —AlexSm 02:13, 25 April 2008 (UTC)

Well I'll be a son of a.. I can't believe it was right in front of me this whole time.. -- Ned Scott 02:39, 25 April 2008 (UTC)

Booksource

Resolved

 – .js not= .css Franamax (talk) 08:33, 24 April 2008 (UTC)

I am (again) having difficulty with the booksource function, xan someone have a look at my monobook thingy and tell me what I need to do to fix it? Thanks, DuncanHill (talk) 07:34, 24 April 2008 (UTC)

You have put in some stuff to your monobook.js that should be in .css. If you put stylesheet directives into a javascript file, it will fail to load. Back out the 19Apr revision. Franamax (talk) 08:19, 24 April 2008 (UTC)

Does back out mean remove? DuncanHill (talk) 08:23, 24 April 2008 (UTC)

OK, thanks - seems to be working now! DuncanHill (talk) 08:27, 24 April 2008 (UTC)

I'm putting your monobook.js back on my watchlist!! :) Franamax (talk) 08:33, 24 April 2008 (UTC)

Thank you - it may surprise you to know that I am not the most technically-minded of people! DuncanHill (talk) 08:41, 24 April 2008 (UTC)

Stick to what you're good at and don't sweat the rest. There's lots of us here ready to help out! :) Franamax (talk) 08:48, 24 April 2008 (UTC)

Editing not showing up in hist

Resolved

 – Problem fixed itself

Anyone see my last edit here? I don't. asenine ^t/_c\^r 16:06, 24 April 2008 (UTC)

Never mind, the problem fixed itself. Still worrying that it was such a slow updation. asenine ^t/_c\^r 16:07, 24 April 2008 (UTC)

Edit window problem

For the past few days, I have had a problem with my edit window, which when first opened expands to fill the whole screen, so that I am unable to see the sidebar, page tabs, special characters panel or much more. Clicking on Preview resolves this problem, but it's a nuisance to have to do this every time. I am using Firefox 2; if I use IE, the problem does not appear. I've tried to purge and delete the cache, but this did not help. I think the only change I have made to my set-up since things worked properly is to add the refTools gadget -- which doesn't seem to work anyway, so I have removed it.

If anyone can help me work out what is causing this problem, I would be grateful. RolandR (talk) 21:20, 24 April 2008 (UTC)

It may be a problem due to a user script in your monobook.js; wikEd, for instance, messes with the edit box quite a bit. What happens when you disable it (put // at the start of the line that opens it)? (That script jumped out at me because it changes the edit box, and only runs on Firefox.) --ais523 21:48, 24 April 2008 (UTC)

Also, the refTools gadget interferes with wikiEd, so the 2 won't work together (though if you have both installed refTools should just quietly fail). Mr.Z-man 21:56, 24 April 2008 (UTC)

Yes, disabling wikEd does resolve the problem. Which is odd, because I have been using it for a long time without problem. There was a Firefox update last week; but if that was causing the problem, I would expect other editors to have reported it too. So either that is not the problem, or something else in my set-up has caused it to act strangely. Any ideas? I would like to be able to use wikEd. RolandR (talk) 22:09, 24 April 2008 (UTC)

OK, I have resolved it; somehow, wikEd had become set to default open in full screen mode. Thanks for your help in identifying this RolandR (talk) 22:15, 24 April 2008 (UTC)

Images...

How do I upload an image to a page? I have tried image codes, URL, Dirct Links, HTML. I don't see a download button either. Again, how do I upload an image to a page on Wiki? -Gabby (Wanna-be good page editor) —Preceding unsigned comment added by 98.193.216.99 (talk) 21:45, 24 April 2008 (UTC)

You need to create an account first (via Special:CreateAccount); then use the 'Wikipedia:upload file' link in the sidebar on the left to upload the image to the wiki, and finally see Help:Image for information on adding the image to an article. --ais523 21:51, 24 April 2008 (UTC)

MediaWiki Suggest (for lack of a better link)

Neat! I just noticed it. When did it go online? — Bob • (talk) • 01:27, 22 April 2008 (UTC)

About an hour ago. See the Wikitech-l thread. —Simetrical (talk • contribs) 03:03, 22 April 2008 (UTC)

That's the suggest in the search box, correct? Can it be disabled? seresin ( ¡? ) 04:24, 22 April 2008 (UTC)

Put   os_autoload_inputs = []   into your monobook.js. —AlexSm 05:37, 22 April 2008 (UTC)

Be easier to just check the 'disable' box in your preferences. :) --brion (talk) 07:29, 22 April 2008 (UTC)

Ah- that would be My Preferences → Search → Disable AJAX suggestions. --— Gadget850 (Ed) ^talk - 11:05, 22 April 2008 (UTC)

Ah, I love this feature. I've already used the same feature as a Wikipedia search engine plug-in for Firefox for some time now. But now that it is part of the regular search box I will have access to it even when I use other browsers. Thanks!

--David Göthberg (talk) 12:18, 25 April 2008 (UTC)

New search box "Feature"

OK, so when did someone start playing with the search box? I used to use it as a quick store of the list of common pages that I visit and now it tries to provide me with some "suggestions" as to what I want to search for - driving me up the wall! How do I disable this "Feature" that has appeared?? Can I? It is *most* irritating, at least for myself. User A1 (talk) 16:00, 23 April 2008 (UTC)

Most people like it. :) But, alas, you can get either the wiki's suggestions or your browser's search history, but not both at once. You can disable it in your preferences; look in the "Search" tab, there's a checkbox. --brion (talk) 17:15, 23 April 2008 (UTC)

Cheers, was driving me nuts! Couldn't find my way around my own browser anymore! User A1 (talk) 00:50, 24 April 2008 (UTC)

Whew! Thanks from me too! I use it the same way User A1 does, and it was driving me nuts too. I checked my preferences page when I first discovered the change and didn't see a check box; maybe it was added after I looked, or maybe I'm just not too bright. :) In any case, thank you thank you thank you for the info! -- edi (talk) 06:19, 24 April 2008 (UTC)

And once again anons get stuck with no way to disable annoying add-ons --172.170.34.223 (talk) 22:16, 25 April 2008 (UTC)

There's a simple way to solve that ;) _TalkIslander 23:17, 25 April 2008 (UTC)

Image deleted, but description page still there and no entry in deletion log

I just deleted Image:Tailspeace.gif (per the uploader's request on WP:MCQ), and got the usual "Action complete" page. However, going back to look at the image, I see that the description page is still there even though the image no longer exists. Special:Undelete/Image:Tailspeace.gif shows one deleted file revision and no matching entries in the deletion log. What's going on? —Ilmari Karonen (talk) 01:01, 24 April 2008 (UTC)

It might be an intermittent bug, I came across one or two PDF files like this recently. I didn't think much of it, just to tag the pages as "missing image" and move on (I didn't check the deletion log either). But yes, it is strange. MER-C 11:39, 24 April 2008 (UTC)

Deleted. Are you sure you didn't just delete the image and not the page? — Edokter • Talk • 12:52, 24 April 2008 (UTC)

Yes, and anyway there should've been a log entry in any case. —Ilmari Karonen (talk) 20:31, 25 April 2008 (UTC)

Can we rank articles from a list/category by the number of 'what links here'?

As discussed some time ago here, it would be useful if we could determine what are the most popular wikipedia's essays. Do we have a tool that could rank them based on how often they are linked to? The category has over 600 entries, close to 1000 with subcategories, and counting it manually would not be easy.--_{Piotr Konieczny aka Prokonsul Piotrus| talk} 20:26, 24 April 2008 (UTC)

A script could easily scrape the whatlinkshere details from the API, a toolserver query or a database dump. Try asking at WP:BOTREQ or tswiki:Query service. Happy‑melon 11:48, 25 April 2008 (UTC)

Protection history of a page

Is there a way to pull up just the protection/unprotection history of a page? It strikes me that this could be useful in identifying trends in the vandalism of a page, or in reviewing the effectiveness of protection. DuncanHill (talk) 20:49, 24 April 2008 (UTC)

Hit the edit button, then change "edit" in the URL to "protect". That's how admins look at it when they go to apply protection, and I think non-admins can at least see the page. EVula // talk // ☯ // 20:55, 24 April 2008 (UTC)

Stupid me, there's an even easier version: click on the History tab, then click "View logs for this page" under the title. EVula // talk // ☯ // 20:55, 24 April 2008 (UTC)

Many thanks - both work, and are just what I was looking for! DuncanHill (talk) 20:57, 24 April 2008 (UTC)

If (like me) you find two clicks overly strenuous, you could use Wikipedia:WikiProject User scripts/Scripts/Logs link. Algebraist 08:34, 25 April 2008 (UTC)

However, this script also duplicates "Logs" (for user actions) link on user pages, which was added to MediaWiki after the script was created. —AlexSm 12:13, 25 April 2008 (UTC)

Ah, I hadn't noticed that. It can be fixed easily enough, but since I don't know JS I'm not going to do it myself. Algebraist 14:19, 25 April 2008 (UTC)

Yep, the action=protect page is accessible to non-admins; it looks the same as it does to admins except that all the options are read-only and grayed out (because non-admins can't protect or unprotect the page). --ais523 08:38, 25 April 2008 (UTC)

Just a note: as far as I remember, protect log won't show you protections made before the page was moved to the current title. —AlexSm 12:13, 25 April 2008 (UTC)

Rights Log

ok admins can see my rights log with this url [16] but my contributions supposedly using the same method gives me this [17] - you can see the difference immediately, but why do sysops have a right to see "user rights" that a user can see for him or herself without knowing the URL? is this a wiki? I was just shocked to see what had been written about me months after what had happened. This can't be right. BpEps - t@lk 03:37, 25 April 2008 (UTC)

I'm not really sure what you mean here. The first URL you link is a listing of the logged actions done to you; the second link is a listing of the logged actions done by you. The second one is empty because you haven't blocked anyone, deleted or protected a page, etc... Titoxd^{(?!? - cool stuff)} 03:40, 25 April 2008 (UTC)

Please see my response at User talk:Betacommand, any IP or User can see any user's rights log. MBisanz ^talk 03:44, 25 April 2008 (UTC)

This is correct, anyone can see anyone elses 'contributions' to this site (wiki if you will). Try looking here, this is my contributions, note that you can see my contributions, even though you are not an admin. Understand any better now? ← κεηηε∂γ ^{(talk) (secret)} 09:04, 25 April 2008 (UTC)

One of different ways for anybody to see that log is by starting at User:Bpeps and clicking "history" and then "View logs for this page". Maybe there should be even more ways but there is really no conspiracy to keep this secret or hard to get at. PrimeHunter (talk) 11:00, 25 April 2008 (UTC)

The biggest problem is that the first url is for all rights, but the other url is for all log events in the year 1AD or before!! Happy‑melon 11:52, 25 April 2008 (UTC)

That's correct. Try this link: [18] --Kbdank71 17:32, 25 April 2008 (UTC)

Something wrong with diffs

Check this out. It shows only one line has changed, but if you check the history, a user has posted several comments in the intermediate revision. And yet, these comments are not shown in the diff. Strange? Gary King (talk) 06:59, 25 April 2008 (UTC)

My NavPops show it as two comments. And WikEd's extended diff view does it right. But the vanilla diff view shows the modification of one heading. Curious. This, that and the other _[talk] 10:39, 25 April 2008 (UTC)

Faulty diffs are sometimes reported here. I haven't done it in case others want to see the problem, but the error would probably be fixed by purging the diff page by adding &action=purge to the url. PrimeHunter (talk) 11:06, 25 April 2008 (UTC)

It's fixed now. Did someone purge it, or was that naturally-ocurring? Algebraist 14:11, 25 April 2008 (UTC)

Gadget process

What should be the rules for adding user scripts and style hacks as gadgets so that they can be checked in your preferences. Please join an ongoing discussion on Wikipedia_talk:Gadget. Please also consider to add Wikipedia_talk:Gadget and Wikipedia:Gadget/proposals to your watchlist. Сасусlе 13:00, 25 April 2008 (UTC)

RefTools gadget not working

The RefTools gadget appears to have stopped working. Now, when I click on the add citation button, a drop down box with the names of the templates appears, I click on the template I want and then nothing happens. I use Safari 3.1 on WinXP. DuncanHill (talk) 13:34, 25 April 2008 (UTC)

FWIW, it's working fine in Firefox 2.0.0.14 on XP. Algebraist 13:58, 25 April 2008 (UTC)

Seems to work in IE7 (though the button is at the other end of the edit toolbar than I am used to). DuncanHill (talk) 14:14, 25 April 2008 (UTC)

Ah, I was wondering how you got hold of unreleased software... It's not just you; it's broken on my Safari 3.1 too. Algebraist 16:34, 25 April 2008 (UTC)

Yeah, I just realized that, apparently Safari doesn't like assigning an "onclick" attribute to dropdown box items. I'm switching it back to the buttons, you'll have to bypass your browser's cache of MediaWiki:Gadget-refToolbar.js to see the changes. Mr.Z-man 16:55, 25 April 2008 (UTC)

Working now - many thanks, :) DuncanHill (talk) 18:04, 25 April 2008 (UTC)

Diff problem

Why are diffs not showing the actual things in the diffs, as here? Badagnani (talk) 22:54, 25 April 2008 (UTC)

Appears to be related to the thread above. x42bn6 Talk Mess 00:00, 26 April 2008 (UTC)

Another diff problem

This diff is showing a repeated change in places there is no actual change. You can see what it looks like here if you're interested.--uɐɔlnʌɟoʞǝɹɐs 16:49, 25 April 2008 (UTC)

Yeah, this is a known bug we haven't been able to resolve yet. --brion (talk) 17:15, 25 April 2008 (UTC)

Does it have a bugzilla entry? --Erwin85 (talk) 20:48, 25 April 2008 (UTC)

Another oddity [19] Nakon 21:00, 25 April 2008 (UTC)

These seem to be popping up by the squillion just now, according to reports in #wikimedia-tech. Brion says he's enabling some logging to check out what's happening. —Simetrical (talk • contribs) 22:38, 25 April 2008 (UTC)

Ok, think I've got it worked out; looks like it was a low-level library mismatch on the binaries or something. Recompiled the diff extension fresh for each version of Fedora we've got and it seems to have cleared up. --brion (talk) 22:04, 26 April 2008 (UTC)

I want to know my all bots total editcuont

I have 10 interwiki bots.

I want to know total editcount at one time.

How to??

Wybot (talk · contribs · count)

I can know en: count by above user9 template.

but I wnat to know "TOTAL COUNT" in ko:, en:, zh:, ... bots.

Help me, please! ^^

this tool does what you want - it counts edits across all Wikimedia Foundation projects. . Graham87 03:39, 26 April 2008 (UTC)

Thanks very much! :)))) -- WonYong^Talk 04:03, 26 April 2008 (UTC) —Preceding unsigned comment added by WonRyong (talk • contribs)

The tool doesn't appear to count contributions, merely print them, as far as I can tell. Several API queries could determine the amount of edits made across wikis (including deleted edits). Gracenotes^T § 04:23, 26 April 2008 (UTC)

The tool counts the edits near the end of it's report, like this: "749 projects scanned. X contributions found in Y projects." The tool only counts edits that have not been deleted. The edit count of the currently logged-in user is shown in Special:Preferences, so accessing the number of edits through Special:Preferences for the bot account in each wiki and adding them up would be a straightforward (if possibly tedious) way of finding the total edit count including deleted edits. Graham87 15:37, 26 April 2008 (UTC)

What are Main Page/1, etc.?

What are Main Page/1, etc., for? They appear to be copies of the Main Page. And why is the Main Page under the cascading protection of these pages, not its own? This, that and the other _[talk] 10:36, 25 April 2008 (UTC)

It's a protection measure to stop the main page getting vandalised. Unitl recently, the most popular thing to do with an admin account if you cracked its password was to delete the main page. If the page is deleted, any protections on it are lost, so if the main page was deleted, it was fair game: anyone could edit it. The ten mainpage backups had two purposes: firstly, they made the main page salted if it were ever deleted, so it couldn't be recreated with vandalism or graffiti; in order to expose the main page completely, the vandal would have to delete or unprotect all ten of the backups, which would take long enough for them to be desysopped. Secondly, if the main page was deleted, a bit of custom javascript changed the default "this page is salted" message to one saying "the main page is temporarily unavailable, please go to one of the backup pages below", so that users weren't scared away by a non-existent main page. It sounds like something out of a cold-war movie, but at one time we really did need all this continuity-planning to keep the main page penis-free! Nowadays, the main page can't be deleted by anyone (there's an exception hardcoded into the mediawiki software, with a very obscure emergency workaround), making the multiple backups a bit unnecessary. However, in order to remove the cascading protection of the main page templates, a compromised account would still need to change the levels on eleven pages, which would certainly get them noticed. All adds up to why the average lifetime of a compromised admin account is seven minutes :D !! Happy‑melon 12:03, 25 April 2008 (UTC)

It's not that obscure. You think main page-deleting vandals are incapable of obsessively searching talk archives? It's probably still worth having the backups just for the deletion protection, apart from the protection-protection issue. Algebraist 14:09, 25 April 2008 (UTC)

Hahaha, wonderful! Even though I wasn't the one asking: Thanks for that explanation Happy-Melon. That's just like what I tell my students: "When you build security systems you need to use multi-level in-depth defence." and "Even pretty primitive defence lines make life tougher for the attacker. Defence lines add up and sometimes even multiply up." and "A sufficient dose of paranoia makes you a good security engineer. I call it 'risk aware', not 'paranoid'."

--David Göthberg (talk) 14:09, 25 April 2008 (UTC)

I am officially calling for the arrest and paradigmatic punishment of Happy-melon for publicly discussing state secrets. An example should be made of him for compromising Wikipedia security; may his fate be a horrible one—saddened as I am to say this—for it is for the greater good... Waltham, The Duke of 00:29, 26 April 2008 (UTC)

:S Happy‑melon 13:17, 27 April 2008 (UTC)

Category list message

I am planning to add a sentence to the MediaWiki message that goes on top of the list section in category listings. Currently it says for instance:

The following 74 pages are in this category, out of 74 total.

I intend to extend it to say something like this:

The following 74 pages are in this category, out of 74 total. Updates to this list can sometimes be delayed some day.

I am not a native English speaker and I think that sentence should be short and clear, so suggestions for improvements are very welcome.

The reason I want to add this sentence is that MediaWiki runs updates of category listings as a low priority job. Thus when the servers are very busy it can sometimes take several days before a page is added or removed from the category listing, but usually it takes less than an hour. User pages seem to be even lower priority so when normal pages might take an hour user pages can take 4 days. I constantly run into editors who are worried about that pages don't turn up or aren't removed in the category listings when they update pages. They go on and do all kinds of purges, edits and debugging of their pages to try to solve it. So I would like to inform about the delay in some way.

--David Göthberg (talk) 12:24, 25 April 2008 (UTC)

Since no one answered here. I asked around on IRC and people suggested a better message:

The following 74 pages are in this category, out of 74 total. Updates to this list can occasionally be delayed for a few days.

I will add this message now.

--David Göthberg (talk) 14:14, 27 April 2008 (UTC)

Special:PrefixIndex

Where Special:PrefixIndex allows you to search article names that begin with a certain text string, Wikimedia.de grep is a recently improved tool that allows you to search text strings anywhere they appear in the title of any name space. The grep tool also lets you use wildcard characters and other characters (see Regular expression) to formulate your search string. H(ä|ae?)ndel will find "Handel", "Händel", and "Haendel" no matter where they appear in the title. (S|s)chool will find School and school, even if it does not appear at the prefix. Please consider creating Special:RegularExpressionIndex that includes all the features of Wikimedia.de grep. GregManninLB (talk) 15:13, 26 April 2008 (UTC) P.S. I recently discovered my gadgets tab and I have become a big fan. Keep up the good work. GregManninLB (talk) 15:15, 26 April 2008 (UTC)

Let's see, we'd be scanning 5 million rows for a regex. I tried this query on a copy of the database, and found this:

124 rows in set (1 min 22.35 sec)

The reason we can scan for a prefix is because we have a list of pages in alphabetical order sitting in the database, which makes searching much quicker:

88 rows in set (22.92 sec)

So, yes, there is a good reason that we don't allow searching by regexes through the main interface. I hope this answers your query. — Werdna talk 04:54, 27 April 2008 (UTC)

I very much doubt your database takes 22.92 sec for page_namespace=N AND page_title LIKE 'Prefix%'. On the toolserver (which is, as River aptly says, about as fast as a snail that's overdosed on sleeping pills ― plus amidaniel and I are torturing the database with huge statistical queries at the moment) I get:

mysql> SELECT page_title FROM page WHERE page_namespace=0 AND page_title LIKE 'Han%';
+---------------------------------------------------------------------------------------------------------+
| page_title                                                                                              |
+---------------------------------------------------------------------------------------------------------+
| Han                                                                                                     | 
| Han'er                                                                                                  | 
. . .
| Hanābilites                                                                                            | 
| Hančaroŭka                                                                                            | 
+---------------------------------------------------------------------------------------------------------+
7805 rows in set (0.03 sec)

The issue isn't that they're clustered together, it's that it doesn't need to scan anything at all. It just jumps right to the correct row, using the index. You can't do that for non-prefixed string searches. I'm guessing you didn't specify the namespace, and MySQL was too stupid to do a union over all the namespaces, so it just scanned the table rows.

It would be perfectly possible to allow things like regex search, anyway, but a proper setup is needed. gmaxwell once remarked that he got working regex searches of not just the page titles, but the page text, and it worked fine, with maybe ten-second latency but high throughput (since multiple regexes were queued up and all run at once when the last batch finished). But you can't just run a MySQL query with REGEXP and accept sane performance.

For matching of plain words within page titles, of course, we already use the ordinary Lucene search ― although there's no way that I know of to just search the titles and not the page text as well. —Simetrical (talk • contribs) 02:25, 28 April 2008 (UTC)

Simplied upload system

I want to simply upload system on slovenian Wikipedia (sl.wiki.x.io). My idea was firstly to have similar upload wizard that here on en: and then not standard input box with preloaded information template syntax, but sometihing like this: sl:Uporabnik:Mihael Simonič/Peskovnik 5. But there is problem. MediaWiki allows you only to use imputbox with following types: "create", "comment", "search" or "search2". So how to do: simple select box with options for licence, only one "Potrdi=OK" button and "generate code" or "preview code" and at the end: Upload button? For example see this site and try imagine that exercises are my upload wizard, inputboxes are like mine for Opis, Vir, Datum, Avtor, Dovoljenje and Druge različice, button Check is for Upload and button Show answer is Preview code.

I try do this in many ways but it seems that MediaWiki don't allow all HTML syntaxes. Greetings, --Smihael (talk) 08:32, 27 April 2008 (UTC)

It might be a better idea to ask at the Slovenian equivalent of the village pump...... Dendodge.Talk^Help 12:10, 27 April 2008 (UTC)

I ask there, but no reply yet. If you want I can translate my mockup. --Smihael (talk) 12:23, 27 April 2008 (UTC)

English translation. --Smihael (talk) 12:36, 27 April 2008 (UTC)

add section button

Is there any way to make sure that some text always stays at the bottom of a page even if someone adds a new section with the add section button? Thanks. 64.231.93.11 (talk) 17:01, 27 April 2008 (UTC)

Yes— see {{UserTalkReplyhere}} for an example. --— Gadget850 (Ed) ^talk - 17:05, 27 April 2008 (UTC)

Or, you can do it with this:

<div style="position:fixed; right:0; bottom:0; display:block;">Text at bottom!</div>

--— Gadget850 (Ed) ^talk - 20:18, 27 April 2008 (UTC)

You mean, let's forget about IE users? —AlexSm 20:38, 27 April 2008 (UTC)

I'm presuming you mean that position:fixed does not work for IE? What is the alternative, then? --— Gadget850 (Ed) ^talk - 20:43, 27 April 2008 (UTC)

Seems to work with IE7. --— Gadget850 (Ed) ^talk - 20:48, 27 April 2008 (UTC)

Doesn't work with IE6 for sure. With IE7 support is contitional (see Comparison of layout engines (CSS)), so you have to test it on Wikipedia pages. —AlexSm 21:05, 27 April 2008 (UTC)

Learn something new every day. I might actually learn CSS rather than hacking it. --— Gadget850 (Ed) ^talk - 00:56, 28 April 2008 (UTC)

Surely you mean "position: absolute", given the original request. —Simetrical (talk • contribs) 02:26, 28 April 2008 (UTC)

Depends on what you want. Fixed puts it at the bottom of the visible window where absolute puts it at the bottom of the page. See User:Gadget850/Sandbox1

The original request made it sound very much as if he wanted it to be like ordinary text on the bottom of the page, which would mean absolute positioning, not fixed. I know the difference, yes. —Simetrical (talk • contribs) 03:54, 28 April 2008 (UTC)

ParserFunctions help

Hi, I'm trying to use a nested #if for logic in {{Infobox netballer}}: If variable previousname exists, then display one of two options dependent on the existence of namechangedate. I have 3 test pages:

• Belinda Colling - no name change - template vars are being displayed when they shouldn't be.
• Adine Wilson - name change with date - displaying correctly
• Anna Stanley (name cahnge with no date) - being displayed incorrectly.

What have I done wrong?? dramatic (talk) 09:57, 28 April 2008 (UTC)

Hi Dramatic, I fixed the problem for you I think. When you use a parameter in the conditional part of a parser function to detect if it exists, you need to place a pipe after the parameter name: {{#if:{{{param|}}}|Do this|Else do this}}. A pipe in a parameter gives that parameter a default value if the parameter is not specified. In this case, if the parameter {{{namechangedata|}}} is blank, then it will default to null, causing the "else" part of the if statement to execute; without the pipe, the parameter defaults to {{{namechangedata}}}, which prevents the "else" from ever being executed. You may want to read Help:Parameter default for more information. Let me know if you have other questions on this. --CapitalR (talk) 10:06, 28 April 2008 (UTC)

Thanks! Yep - I understand that and have just tweaked the bolding. Looking good now. dramatic (talk) 10:14, 28 April 2008 (UTC)

Error message

Whenever I edit Wikipedia:Requested articles/Natural sciences/Biology I get a Wikimedia Foundation error message. Any idea what's going on? It's a very large page, so I presume that might be part of the explanation. The funny thing is that the edits I make actually go through, they just appear not to. Richard001 (talk) 11:34, 28 April 2008 (UTC)

Edit? I had trouble getting that page to load! My browser gave me a pop-up message saying it was having trouble with a script that is likely to cause my computer to become unresponsive, and do I want to abort the script. I said yes, and the load completed. I'm not even going to try to edit. Why the heck does this page have multiple huge lists in it? Look how the last list is broken up: Ac to Ag, Al to Am, etc. It is obvious each section was intended as a separate index page, and maybe it was at one time. Someone please break up the page – if you can! --A Knight Who Says Ni (talk) 14:51, 28 April 2008 (UTC)

You get the foundation error message if you try to edit a heavily-transcluded template too. I think the first thing the software does is record that the edit has taken place, then it starts updating all the things that have to be updated; in the case of a template, it has to compile a huge number of job queue requests, and for a big page, it has to write a lot of data to various tables. I assume there's a very low-level test somewhere which, if the servers don't reply with anything useful after a certain time, assumes something's gone wrong and loads the error message. In fact, everything's working fine, it's just taking a while to sort everything out. I assume if you tried to delete a page with 4999 revisions, or rename a prolific user, you'd get the same issue. Happy‑melon 15:27, 28 April 2008 (UTC)

There are only five templates used in the page as shown after the edit window. {{search}} is used 92 times and {{missing article}} is used 1092 times. --— Gadget850 (Ed) ^talk - 17:02, 28 April 2008 (UTC)

Displaying MediaWiki:Talkpagetext outside of talk namespaces

Clicking the "edit this page" tab on a Talk page causes the text in MediaWiki:Talkpagetext to display above the edit window. This is helpful because many new users do not otherwise know to sign their posts. Unfortunately, the Help desk is not a true talk page, because it is in the Wikipedia: (Project:) namespace, and thus MediaWiki:Talkpagetext does not display above the edit window when a user enters a new question. Not surprisingly, the Help desk gets a lot of unsigned questions. We display an instruction to sign one's posts, but it's up in the Wikipedia:Help desk/Header, mingled with many other instructions, and is therefore not visible while the user is editing. The Reference desk and other talk-like pages have the same problem; I even see some unsigned comments on this page. Is there a way we can enable MediaWiki:Talkpagetext to display above the edit window on a per-page basis, for pages outside of any talk namespace? --Teratornis (talk) 15:42, 28 April 2008 (UTC)

I never noticed that. There must be another MediaWiki setting for that. It won't fix where a comment is added by editing the last comment, but it can't hurt and will probably help. I had my head in that recently, so I will poke around a bit. Someone will probably beat me to it. --— Gadget850 (Ed) ^talk - 16:04, 28 April 2008 (UTC)

I believe MediaWiki:Talkpagetext displays above the edit window regardless of how the user opens the edit window on a true talk page. Also, it seems I was incorrect when I said Wikipedia:Help desk/Header has an instruction to sign one's Help desk posts - the instruction is not there now. The Help desk instructions undergo cyclical expansions and contractions, as various people add instructions in the hope that some users will read them, and then other people decide instructions are bad. --Teratornis (talk) 16:12, 28 April 2008 (UTC)

If you start a new section and look at the browser page source, search for "<!-- start content -->". On an article talk page, the next line starts with "<div id="talkpagetext" class="plainlinks metadata". On a non-article talk page, that line is blank, so it never calls a message. --— Gadget850 (Ed) ^talk - 16:18, 28 April 2008 (UTC)

I always found it very inconvenient that "Wikipedia:" namespace is used for discussions although it is even (#4) namespace. This creates difficulties for system messages (like the one described above), for bots and userscripts. In another Wikipedia we have a piece of JavaScript in Common.js that appends the reminder to sign just under the edit window (MediaWiki:Copyrightwarning), it works in all odd namespaces and it uses regexp to cover most Wikipedia: discussion pages; but I guess that's not an option for English Wikipedia simply because there are too many Wikipedia: discussion pages. —AlexSm 16:28, 28 April 2008 (UTC)

Page can't be displayed

Page Wikipedia:WikiProject Stub sorting/Stub types doesn't seem to display. Do others get the same issue? Perhaps there is a technical fault? 16:51, 28 April 2008 (UTC)

Does not display for me on Safari 3.1 on WinXP (I just get a white screen). DuncanHill (talk) 16:54, 28 April 2008 (UTC)

Yes, it doesn't display. I removed the Geography transclusion and now it does. My guess is that the page is too large somehow; I ran into a similar issue elsewhere earlier this week. The easiest fix, if I am right about the problem, is to split the page into several subpages. — Carl (CBM · talk) 17:00, 28 April 2008 (UTC)

This was also discussed at Wikipedia:Village pump (technical)/Archive 34#WikiProject_Stub_sorting/Stub_types where it loaded for some and not for others. PrimeHunter (talk) 17:41, 28 April 2008 (UTC)

I don't think it is browser-specific. Using cURL, attempting to load the page retrieves 0 bytes before it times out (~15 seconds, apparently on the server-side – see below). Using Firefox, and cURL, this diff (where the newer page does not have the geograpy section) loads. However, this diff does not load (the newer page has the geography section, and is thus larger). In the case where the page loads, cURL reports 0 bytes received for about 10 seconds before reading all 863k in the next 2 or 3 seconds. In the other case, I attempted to set cURL's timeout to 60 seconds; but, the server still closed the connection after 14~15 seconds. That's all probably a bunch of useless analysis; but, it may help solve whatever trouble is going on. Neier (talk) 23:22, 28 April 2008 (UTC)

The problem is that the page is too big. I double-checked that with the tech channel on IRC, who can look at the error logs. They also say that, in general, if a very large page comes back blank, it is almost certainly because the page is too big, so there's no need to go there and ask :).

The issue is that each Mediawiki instance is limited to 80mb of memory, and if this is exceeded all that is returned is a blank page. The error is recorded on the server when this happens. The structure of the page makes a big difference in how quickly this limit is reached, so a 1MB page of text with no links may render while a 500KB page with lots of complicated structure may run out of memory.

The only solution is to make the page smaller. And we should do that; it's no good to have broken pages that cannot be rendered. — Carl (CBM · talk) 00:04, 29 April 2008 (UTC)

The page is unpredictable - sometimes it loads and sometimes it doesn't. At the moment it's causing quite a bit of trouble because the servers in general; are struggling (notice how long it takes for categories to update these days?). For that reason, there's a link-only form at Wikipedia:WikiProject Stub sorting/List of stubs, which does pretty much the same job. The important thing is where possible to never edit the "stub types" page directly, since it takes a near superhuman effort to get it back the way it should be (which is still used by some when it works) - it took Neier nearly an entire day recently to fix the page when someone removed a section from it. And given that it is still used and causes so much trouble to fix up, can I ask for any changes to it (such as turning it into links only) be debated on its talk page before anyone takes any unilateral action like removing parts of it? Untilo then, if you have any problems with it, do what the rest of us do and use Wikipedia:WikiProject Stub sorting/List of stubs. Grutness...wha? 01:37, 29 April 2008 (UTC)

The page just needs to be split into two parts. Yes, that's less convenient, but our servers have practical limits, and this page is past them. If it is failing often enough that people aren't surprised when it's broken - then it needs to be fixed. — Carl (CBM · talk) 01:59, 29 April 2008 (UTC)

CSS Targets

How is the CSS :target selector put to use here? I believe that is what is used to highlight any selected reference among the list of references, isn't it? Can the same selector be used for other purposes manually? Specifically, I have this scenario in mind: In the 2008 Indian Premier League article, the Played column in the Standings table will link to a certain anchor in the Results table, representing a particular cell. Now when the anchor is navigated to, the :target selector comes into play and highlights the cell (div). Is that achievable? That brings up another question: can there be multiple instances of the same anchor, so that all those cells that have the anchor are all highlighted? I have very limited knowledge about CSS-based presentational techniques, any help is more than appreciated. :-) --soum ^talk 19:15, 28 April 2008 (UTC)

search bar

Hello Wiki,

I am a big fan of your wikipedia web tool. I use it half a dozen times a day. It’s an invaluable resource.

However, I have one minor complaint. As someone who spends most of his working hours on the computer, mouse clicks have become a commodity. If I can find ways to reduce the number of clicks to complete a task I will.

Again, this complaint is so minor, but it has a ton of significance to me. Here’s the scenario….when I launch your wikipedia page, I have to mouse over the text entry box, “mouse click” to drop my cursor into the box, and then I can type my search text.

It’s again such a minor annoyance, but it makes a big difference in time conservation and ease of use. I’m imagining there is simply line of code that could be added that would automatically put the viewer’s cursor into the dialogue box.

If that was done, I would have to say wikipedia is flawless.

Please let me know if you can accommodate me.

Yours,

Leo. —Preceding unsigned comment added by 96.233.182.141 (talk) 19:18, 28 April 2008 (UTC)

This has come up before, in various other places. We generally try to avoid second-guessing what users want to do where possible, and so I doubt that this would ever be implemented. For instance, if the cursor were placed automatically in the search box, it would be impossible to scroll the page using a mouse wheel without clicking outside the search box to remove the cursor focus, which would be an inconvenience for many users. However, what has been done in various places (Special:Userlogin, for instance) is to place the search box as the 'top focus', so pressing the tab key once will shift the focus to the entry field. There's no reason why this couldn't be done, although of course it would conflict with said other focuses. Happy‑melon 20:05, 28 April 2008 (UTC)

If you create an account you can set your preferences to automatically set focus to the Search box, otherwise just press the Tab key instead of mousing and clicking in the Search box. Sbowers3 (talk) 20:13, 28 April 2008 (UTC)

Generally speaking, I can just hit tab in my browser and it takes me to the first text field, which is the search box. No mouse required. EVula // talk // ☯ // 20:29, 28 April 2008 (UTC)

Wikipedia:Articles for deletion/David Sustak

Can somebody explain to me why the link at Wikipedia:Articles for deletion/David Sustak is showing up as a red link on the David Sustak article? The code looks okay, and when I click on the pre-loaded debate link, it goes to the correct page. Corvus cornix_talk 21:03, 28 April 2008 (UTC)

Happens sometimes. This can be fixed by purging and/or simply saving the article twice after the AfD (no edit required). – sgeureka ^t•c 21:16, 28 April 2008 (UTC)

Change to default image page template ("Licensing" -> "Licensing:")

It would seem to me that recently uploaded images have a section title of "Licensing:" instead of "Licensing" (note the colon in the new version). Was this a deliberate change? It looks like it happened in this change (but I am not positive) about a week ago. Anyone know what's going on? - AWeenieMan (talk) 04:48, 14 March 2008 (UTC)

This has been fixed in the repo. May take some time before it is live here. AmiDaniel (talk) 06:02, 15 March 2008 (UTC)

I brought this up a while back and AmiDaniel said it was fixed, but the software still appears to be adding a colon onto the word "Licensing." - AWeenieMan (talk) 21:37, 28 April 2008 (UTC)

Monobook Question

Can anyone tell me why

importScript('User:ais523/editcount.js'); //[[User:ais523/editcount.js]]


importScript('User:Ais523/hidetopcontrib.js');

isn't working in my monobook but is working in my sock's monobook-- penubag  (talk) 02:53, 29 April 2008 (UTC)

Hmm, weird, if I subst it, it works now...-- penubag  (talk) 04:57, 29 April 2008 (UTC)

Maybe there were stray invisible characters, or something like that. (I don't know what happens if you put non-breaking spaces in JavaScript source, for instance.) Also, it's possible you forgot to bypass your cache first time. --ais523 18:41, 1 May 2008 (UTC)

Are WP:AN and WP:ANI broken for anyone else?

All that appears on those pages (even when I try to edit them) is WP:VANDAL. I've purged and refreshed. shoy 03:15, 29 April 2008 (UTC)

It happened to me too, but they just fixed themselves a minute ago. seresin ( ¡? ) 03:18, 29 April 2008 (UTC)

It's currently happening for me on ANI (AN seems to be fine)... --Umrguy42 (talk) 04:27, 29 April 2008 (UTC)

Someone vandalized one of the templates on the page and caused all the text to be white. The template is now protected and the user is blocked.¤~Persian Poet Gal ^(talk) 04:29, 29 April 2008 (UTC)

PPG, I'm still getting the same thing, even after clearing my cache. (Well, hitting CTRL+F5 in IE7, which I believe is supposed to do that, anyway.) --Umrguy42 (talk) 04:32, 29 April 2008 (UTC)

No it is a vandal everyone...(ec)Ugh not again! Anyone who sees this message, track the templates that are posted on the AIV/ANI/AN and see if there are any suspicious changes.¤~Persian Poet Gal ^(talk) 04:33, 29 April 2008 (UTC)

Person caught, this was the template. Keep your eyes peeled if you see it again.¤~Persian Poet Gal ^(talk) 04:36, 29 April 2008 (UTC)

All of the color templates are now protected. Nakon 04:40, 29 April 2008 (UTC)

Something to think about. Is there a way to see Special:UnwatchedPages for templates only? I'm curious which if any important templates aren't being watched and/or protected. -- Ricky81682 (talk) 05:40, 29 April 2008 (UTC)

StatusChanger does not appear to be working

Hi. I recently implemented the Status changer into my monobook.js . However, when I put it on my userpage, it just says: Out. No "This user is:", no coloured text, and none of the bolding and changes I added. What's going on? Even when I clicked "in" and "busy" on the top bar, which works, the thing still said "Out". Also, it did not form a box, and did not appear on the right-hand corner. The page said it reportedly did not work using IE7. I don't know which version of IE I'm using, but is this the part that doesn't work? Why doesn't the page just transclude onto my userpage? I think the script is just to add the bottons on the top, so the transclusion should work independantly of the script. However, it doesn't work. Is it displayed as "Out" on your browser, too? If the problem is in the transclusion, can it be fixed? Or, is it my browser and should I just remove the tool? Thanks. ~AH1^(TCU) 23:12, 28 April 2008 (UTC)

Hi. Ok, nevermind, I figured it out. It turns out that I was using User:Example's template rather than mine, OOPS! It works now. Thanks. ~AH1^(TCU) 17:22, 29 April 2008 (UTC)

inflation-adjusted conversions

I have no idea how something like this would be done, but is it feasible to have some kind of code that people can embed into the articles that converts a value of a specified currency in a given year into a value in another year? Many articles in wikipedia list currency values like "300,000 in 1975 US dollars" which is practically meaningless unless one was alive then or is an economist/historian with intuition for the relative value of a dollar then. I have seen quite a few cases where the article lists an old value of currency like $300,00 (1975) and then clarifies it afterward with something like $500,000 (2003). That clarification was great in 2003, but as time goes on someone will need to continually update that value for people to be able to make comparisons.

I'm not an economist, but I believe what I am asking about is trying to have automatic conversion of nominal values into real values using a base year that keeps advancing each year. I know there is stuff in wikipedia that kind of does stuff like this: in person infoboxes you can write the code {death date and age |2000|12|31|1900|1|1} and it automatically calculates and displays the age of the person. It would be very functional to be able to do something like {{inflation conversion|USD|1975|300000)} which could display the equivalent value of 300,000 1975 USD in the current year. I guess the conversion factors would have to be generated for each year, but I think this would be trivial. Obviously, there would have to be limits on the dates one can use for any given currency (USD in a 134 AD base year doesn't make any sense). Any thoughts? Dwr12 (talk) 01:37, 29 April 2008 (UTC)

Template:US Inflation gives a CPI. For example: {{US Inflation|Year=1906}} gives {{US Inflation|Year=1906}}. One could simply ${{formatnum:{{#expr: {{US Inflation|Year=1906}} * 300000}}}} to get $ (or make a template to do same). --Splarka (rant) 07:29, 29 April 2008 (UTC)

Discussion template error

{{Discussion top}}, used to archive discussions, fails to display the closing reason if diffs/external links are used. See Template talk:Discussion top. Thanks! Vassyana (talk) 11:40, 29 April 2008 (UTC)

It's not a link problem, it's an equals sign problem. You need to use {{discussion top|1=comment here}}, otherwise you end up uselessly assigning a value to a non-existent parameter with a ridiculously long name, and leaving the actual parameter undefined. Algebraist 15:48, 29 April 2008 (UTC)

Adding symbols to TeX

How does one go about adding symbols or commands to TeX/wikimarkup? For instance, there is a section in Help:Displaying a formula for "Unsorted new stuff": How does one create "new stuff"——if it has to be approved, where does one submit it (e.g., if I wanted to create the character "\grethel", which equals "o\!\varepsilon", $o\!\varepsilon \,\!$, "Greek ethel")? I don't think something like this qualifies as a "feature request", since it is just adding another character to TeX, not some new feature to MediaWiki. ~Kaimbridge~15:49, 29 April 2008 (UTC)

The TeX system on Mediawiki is texvc; adding new commands to our TeX environment would require patching and recompiling the texvc program. To request the change you would need to file an "enhancement" bug in https://bugzilla.mediawiki.org. Select the "MediaWiki extensions" product and then you will be able to select texvc as the "component". — Carl (CBM · talk) 15:58, 29 April 2008 (UTC)

The speaker icon in Template:Audio-pipe

It's come to my attention that some IE7 browsers (including mine) do not display a speaker icon which is (supposedly) rendered by Template:Audio-pipe. The wikitext of this Template contains ⁽ⁱ⁾ , which is rendered correctly on my browser as ⁽ⁱ⁾. No icon is displayed.

Does anyone know anything about this issue? How can I get to see this icon? For background, please see Help:IPA and User talk:Kwamikagami#Non-existent speaker icon. —Preceding unsigned comment added by Wahrmund (talk • contribs) 22:25, 29 April 2008 (UTC) Morris K. (talk) 22:26, 29 April 2008 (UTC)

Accessibility of Wikipedia servers' logs

I'm quite convinced that useful information could be extracted from the wikipedia http servers' log files. In particular, I'd be interesting to do some clickstream analysis to find out the relative importance of the internal links, as this could help to automatically extract some semantic info from the pages. Is there any way to get (at least a sample) of the page access logs? Sorry, I'm not sure this is the right place to ask the question but I didn't find a better one :-) --Ronchet (talk) 17:18, 28 April 2008 (UTC)

Sorry, our logs are unavailable due to privacy issues. MaxSem^{(Han shot first!)} 18:32, 28 April 2008 (UTC)

Indeed, while a proposal to generate information from the logs may gain motion, direct access to the logs would be an unacceptable privacy violation. _(1 == 2)^Until 18:37, 28 April 2008 (UTC)

Of course I do understand the privacy issue, but logs could be cleaned making then anonymous while preserving the other info. Actually, for the scope I was mentioning, only the referrer page would be needed - and only for those records for which the referrer page is internal to wikipedia. --Ronchet (talk) 19:22, 28 April 2008 (UTC)

Of course the logs could be cleaned, but who's going to do it? Our paid devs are worked off their feet just keeping this behemoth on its feet, the volunteer devs are just that, volunteers, and anyone else would face the same privacy issues. Happy‑melon 20:07, 28 April 2008 (UTC)

Anonymizing the logs is not enough; see for instance the AOL search data scandal and the Netflix dataset. --cesarb (talk) 12:24, 30 April 2008 (UTC)

It depends on how well they're anonymized. If IP addresses, internal referers, etc. are deleted completely, so that there's no connection between subsequent requests, there's probably no issue. Of course, this makes them significantly less useful. —Simetrical (talk • contribs) 14:57, 30 April 2008 (UTC)

We've made some stats based off Squid logs available for months now. (The actual HTTP servers don't receive most requests, so their logs aren't useful for this kind of thing.) If I recall the announcements correctly, some type of sampled logs are supposed to have been available dating back several months, and the full log stream should be available for real-time use. But Google doesn't seem to tell me where they are, so maybe I'm just misremembering completely. http://dammit.lt/wikistats/ provides some info, at least, although probably not enough to be useful to you (no referers, etc.). I'm not sure, but my impression is that most of the data you'd find in a typical access log simply isn't recorded at present, although things might be better now than they were several months ago. —Simetrical (talk • contribs) 00:05, 29 April 2008 (UTC)

You perhaps are thinking of this vandalism study. --cesarb (talk) 12:24, 30 April 2008 (UTC)

I don't think so. What relevance does that have to what I said? It does mention the 10% sampled HTTP log I mentioned in passing, though, which helps to convince me I didn't make it up, but I first saw discussion of this on Wikitech-l, before it was implemented. —Simetrical (talk • contribs) 14:57, 30 April 2008 (UTC)

Newpages

I want the "search for new articles by [user]" box to return to Special:Newpages. A box where you could insert the name of a user, and search for new articles started by that user only. Punkmorten (talk) 15:54, 29 April 2008 (UTC)

If this doesn't happen, you can use this tool (at least while it's working). Algebraist 17:45, 29 April 2008 (UTC)

This has been disabled until the necessary data table (which is present on en.wiki but not on most other wikimedia wikis) has been created and populated for all wikimedia wikis (previously each wiki that didn't have the table had to implement a nasty hack to stop the field from displaying and making a mess). It will be back eventually. Happy‑melon 20:43, 30 April 2008 (UTC)