Computing desk
< October 2	<< Sep \| October \| Nov >>	October 4 >

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

October 3

Space efficient index of a list

When you have several indexes, can you combine them to make them more efficient in terms of space?

If you wanted, for example, to access a list of names through 1st name or surname. Do you have to create necessarily one index for each? Intuitively, it makes sense to believe that you'll need 1 index for each way of accessing your information, but that wouldn't be a proof.

Speculative, I wonder whether there is some kind of data structure or hash value that could point to the right entry, when fed one or the other (1st name or surname). — Preceding unsigned comment added by Qlearn (talk • contribs) 00:49, 3 October 2018 (UTC)[reply]

I think the idea of having a general hash that could be generated by either of two seperate records is impossible. I expect it is a lot less efficient than having two indexes but of course you could use the LIKE function; name "John Smith" and search "WHERE NAME LIKE '%John%';" or "WHERE NAME LIKE '%smith%'"; -- Q Chris (talk) 11:09, 3 October 2018 (UTC)[reply]

You could always compress the string. The gain would be minimal and in the practice you don't want to trade processing speed for being stored efficiently. --Doroletho (talk) 13:00, 3 October 2018 (UTC)[reply]

You can implement the index as a trie, in which case you have automatic prefix sharing and performance independent of the number of indexed terms[*]. Then at the end node of each word have pointers to the various rows, each labeled with the target row. [*] Except that in theory the number of entries is exponentially bounded by the depth of the trie, and worst-case search is linear in that depth. --Stephan Schulz (talk) 15:47, 6 October 2018 (UTC)[reply]

Windows installer

Windows installer works great![1] --Guy Macon (talk) 07:11, 3 October 2018 (UTC)[reply]

Very funny. And what's your question? --Doroletho (talk) 11:00, 3 October 2018 (UTC)[reply]

If an ip user posted this they would be reverted for trolling. — Preceding unsigned comment added by 186.148.165.180 (talk) 11:46, 3 October 2018 (UTC)[reply]

I am sorry if that has been your experience. IPs should be treated the same as registered users, and I have seen editors get warned for treating IPs differently from registered editors -- and blocked for repeat offenses. All editors are required to obey WP:TPOC. A small amount of humor is allowed on Wikipedia outside of our actual articles:, see Wikipedia:Humor#Humor outside of articles. We have help for those who have trouble telling the difference between humor and trolling at meta:What is a troll?, along with an article on the concept at Internet troll. --Guy Macon (talk) 12:55, 3 October 2018 (UTC)[reply]

If you are one of those linux evangelists poking fun at Windows, I'd like to say that Windows is easier to install than Linux and on the top of that has improved several aspects in the last years. --Doroletho (talk) 13:07, 3 October 2018 (UTC)[reply]

I am not a Linux evangelist poking fun at Windows. I use both and have found that Windows 10 is somewhat easier to install than Ubuntu and much easier to install than Slackware (Motto: "Slackware is user friendly. It's just picky about who its friends are"). Please don't try to read some hidden meaning into a lighthearted joke. --Guy Macon (talk) 13:44, 3 October 2018 (UTC)[reply]

but like what is your question even lol ~Helicopter Llama~ 14:54, 3 October 2018 (UTC)[reply]

I apologize. I somehow failed to realize that we are on Jeopardy!, where the rules say that you must must phrase your response in the form of question. (the presence of Alex Trebek should have clued me in).

When performing a Windows installation,[2] are hard hats required? Is there a problem with the average house having heavy things falling on your head that I was unaware of? And shouldn't the workers pictured also be wearing orange safety vests? --Guy Macon (talk) 15:31, 3 October 2018 (UTC)[reply]

Typical. Two people loudly insist that I make a comment in the form of a question, then neither of them attempts to actually answer the question... (note to the humor impaired; please attempt to understand the concept of "a joke".) --Guy Macon (talk) 16:03, 4 October 2018 (UTC)[reply]

Weird Excel table behaviour

There's a very convoluted story that I'm trying to avoid having to tell. :) The short version is that I have a series of defined tables, each on their own tab. They each do a series of calculations based on a complicated SUMIFS formula to help me present a variety of different bits of information, all drawn from a data dump (also on a separate tab). When I first created the formulas they naturally ended up with negative values. This was totally expected and the formulas are doing their thing properly. However, to make graphing easier to read, it was later decided to invert the values by multiplying them by -1. All good so far. However, when I go to add another week, the table "forgets" the new formula and the new row shows as negatives. So, assuming a scenario where each of the following weeks should show as "10", my table results in this:

Fiscal Week	Cases credited
W01	10
W02	10
W03	-10

This forces me to fix each row manually by grabbing the previous week and dragging it down to the next row. It's very tedious. How can I gain access to see what Excel "thinks" the formula should be and change it? Even though every cell in every column shows as (complicated SUMIFS formula)*-1, subsequent new rows all forget to add the "*-1". I've drawn three weeks and a single column in here as an example, but in reality there are very many more so I would prefer not to have to redo all of them from scratch. Any help would be appreciated. Matt Deres (talk) 17:46, 3 October 2018 (UTC)[reply]

If the data is correct and the answer is "wrong" (= undesired) then the formula is the thing to fix. It sounds like you need to add to your formula one or both of these functions that are available in Excel:

SIGN Returns the sign (+1, -1 or 0) of a supplied number

ABS Returns the Absolute value of a supplied number

DroneB (talk) 19:01, 3 October 2018 (UTC)[reply]

Maybe I did not explain myself properly. If you were to examine the file right now, every single formula would be correct and giving the proper value. The problem develops when I add the next week in. In the example above, if I enter "W04" in the appropriate cell, Excel automatically extends the formulas in all the other columns - it's a standard function of a formally defined table. The problem is that the automatically created formulas are the old ones with negative values. I have to highlight the row above and manually drag down the updated formulas myself to correct it. Excel is somehow remembering the original formulas I used rather than simply extending the current ones. Matt Deres (talk) 12:36, 4 October 2018 (UTC)[reply]

I'm really not knowledgeable about Excel (or any spreadsheet), but have you tried to delete the row and then re-insert it with the new formula? --Stephan Schulz (talk) 09:48, 6 October 2018 (UTC)[reply]

Yes. No luck. Matt Deres (talk) 16:49, 7 October 2018 (UTC)[reply]

The answer by tjejojyj looks like a hackish workaround that might work [3] Nil Einne (talk) 01:45, 8 October 2018 (UTC)[reply]

While a bit different, it's also possible the answers here may work, especially the later ones [4] Nil Einne (talk) 01:49, 8 October 2018 (UTC)[reply]

Wow - those really look on point - thank you! I won't have access to the file again until next week, but I will definitely give those a try! Thanks again! Matt Deres (talk) 02:26, 9 October 2018 (UTC)[reply]

will there be 256-bit computers?

AWNSER THA QUASTON. — Preceding unsigned comment added by A Loaf Of Bread 2 (talk • contribs) 20:13, 3 October 2018 (UTC)[reply]

See the top of the page "We don't answer requests for opinions, predictions or debate." Dmcq (talk) 20:22, 3 October 2018 (UTC)[reply]

In this case, the "requests for predictions" is a lot like asking "will men ever walk on the moon?" The 2002 Data-Intensive Architecture (DIVA) system was the first 256-bit processor that I know of,[5] the Transmeta Efficeon TM8300 256-bit Processor was released in 2004, and of course there is the AVX-512 and the Nvidia Quadro FX5800. Of course you can always play a game on No True Scotsman if you want to deny that 265-bit processors exist... --Guy Macon (talk) 21:45, 3 October 2018 (UTC)[reply]

I deny that 265-bit processors exist. PrimeHunter (talk) 22:29, 3 October 2018 (UTC)[reply]

There are some google resutls about it: "265-bit+processor". --Qlearn (talk) 11:13, 4 October 2018 (UTC)[reply]

PrimeHunter, are you claiming that the 256-bit Data-Intensive Architecture (DIVA) system and the Transmeta Efficeon TM8300 256-bit Processor didn't happen, or are you claiming that the sources that say that they were 256-bit processors were telling a lie? --Guy Macon (talk) 16:18, 4 October 2018 (UTC)[reply]

I bolded 265 (not 256) to make fun of a typo in your first post. Maybe it was too subtle. PrimeHunter (talk) 16:23, 4 October 2018 (UTC)[reply]

WHOOOSH! (that's the sound of a great joke going right over my head) Thanks! But is it true that there are no 265-bit processors? (Motto: "for when having more addresses than there are atoms in the universe just isn't enough".) I could make one if I had the chips -- see Bit slicing -- but I don't think that anyone is making a 1-bit bitslice processor any more. Hmmm. It might be possible to do it with a FPGA...

I assumed it was a subtle reference to Moon-landing deniers. 216.59.42.36 (talk) 17:36, 4 October 2018 (UTC)[reply]

At least for me, all of those seem to be probably copied typo references to some Mac Pro with a Xenon from 2013, I presume they meant 256 bit due to the Advanced Vector Extensions. Nil Einne (talk) 10:52, 5 October 2018 (UTC)[reply]

A main motivation behind larger bit numbers was the larger address space. However at 64-bit the address is practically unlimited. So, now there is no compelling reason to go for 128-bit or wider architectures, at least in consumer electronics. Ruslik_Zero 18:11, 4 October 2018 (UTC)[reply]

I don't think that we will ever need more than 64-bit addresses. But some of us could really use a larger data path and arithmetic. Bubba73 ^{You talkin' to me?} 18:18, 4 October 2018 (UTC)[reply]

Is this one big enough? [6] :) --Guy Macon (talk) 21:03, 4 October 2018 (UTC)[reply]

That's the good old Motorola MC14500B. Bubba73 ^{You talkin' to me?} 23:05, 4 October 2018 (UTC)[reply]

...A one-bit processor with 16 instructions and an ability -- depending on how you look at it -- to either address one bit of memory or to address any any amount of memory (for the simple reason that it requires an external circuit to handle memory addressing). It is about as simple as a microprocessor can be. Alas, I can't find anyone who still carries them -- even old stock.

http://www.ganssle.com/articles/quirkychips.html has a nice description, and https://www.youtube.com/watch?v=CIs9IuobcPE shows one that was built as a redstone circuit in Minecraft. --Guy Macon (talk) 10:23, 5 October 2018 (UTC)[reply]