Wikipedia talk:Featured article candidates/Citation templates

This page is for discussing the issue of using citation templates for inline citations on FACs. The first proposal (to use only list-defined references where templates are used) failed. A discussion continues about the general problems (if any) caused by templates, and whether a different proposal to discourage their use in FACs, or not to encourage them, is appropriate.

To read the technical points on this page, see Wikipedia talk:Featured article candidates/Citation templates (technical)

First proposal carried over from WT:FAC (rejected)

The proposal was to add the following (underlined) sentence to the end of Criterion 2c.

(c) consistent citations: where required by criterion 1c, consistently formatted inline citations using either footnotes (<ref>Smith 2007, p. 1.</ref>) or Harvard referencing (Smith 2007, p. 1)—see citing sources for suggestions on formatting references; for articles with footnotes, the meta:cite format is recommended. The use of citation templates is not required. If a citation template is used, list-defined references are preferred.

Support. While not barring the use of text-cluttering citation templates, this additional point would at least encourage editors to minimise the amount of clutter within paragraphs that makes editing very difficult to achieve, while making editing and organising the actual reference text easier. Tony (talk) 03:08, 1 February 2010 (UTC)[reply]
Oppose - we should not even be saying which is preferred. I personally, do not like list defined, and it's one step from "preferred" to required. Ealdgyth - Talk 03:13, 1 February 2010 (UTC)[reply]
Oppose. I believe the FA criteria should be a set of requirements for featured articles, and should not be misused as a tool for recommending a certain format to editors. Ucucha 03:15, 1 February 2010 (UTC)[reply]
Oppose per Ealdgyth. Johnbod (talk) 03:23, 1 February 2010 (UTC)[reply]
Oppose — I personally prefer the templates, and I'm starting to use LDR, but I'd never force my choices on others. Maybe someday when the function of the templates is built into the Media Wiki software. Imzadi1979 (talk) 03:31, 1 February 2010 (UTC)[reply]
Comment It may be too early for this type of change; many users are still not aware of LDR. Perhaps we should just mention its existence, via footnote or otherwise. After more editors become familiar with the feature we can survey opinions on whether it is preferred to normal referencing or not. Dabomb87 (talk) 03:55, 1 February 2010 (UTC)[reply]
Oppose - Per instruction creep; if this becomes a policy, a bot can change all of the articles. Until then, let's not bother people who are trying to write good content. Awadewit (talk) 04:00, 1 February 2010 (UTC)[reply]
Oppose per Ucucha. Bishonen | talk 05:04, 1 February 2010 (UTC).[reply]
Oppose. Per Ucucha. Firsfron of Ronchester 05:13, 1 February 2010 (UTC)[reply]
Comment I agree with Dabomb. ~~If this one day becomes part of the FL criteria,~~But if we're advocating LDR, long references should always go at the bottom, templates or not. WFCforLife (talk) 05:58, 1 February 2010 (UTC)[reply]
Oppose Saying it is preferred MEANS it is required, so far as FA reviewerthink. --Wehwalt (talk) 12:55, 1 February 2010 (UTC)[reply]
Oppose: FA criteria should be based solely on the quality of the reader's s experience. Technical considerations that affect only the editing environment or process should not enter into FA consideration. If FA reviewers are convinced that the reader's experience is significantly improved by use of short-form inline citations, with all full-form citations in a bibliography (whatever the section may be called), an FA criterion might express a preference for that system, but it should not have anything to do with whether templates are used. I don't think that criterion would be wise, however, because there are too many article-dependent variables to say that this system is usually better across the board. As for clutter in the edit window—although, again, that should not affect FA evaluation, in my opinion—an article that is very dense with full-form, manually entered, inline citations is more difficult for editors to navigate than an article with, say, one-fourth the number of full-form, inline citations in templates, especially if the templates are vertically filled.—Finell 12:58, 1 February 2010 (UTC)[reply]
- I can imagine cases where it would be reasonable to include editor concerns in the FA consideration. If an author insisted on using (using) HTML codes for every letter in the article (WP:BEANS?), that would impel other editors' ability to edit the article so much that I wouldn't think such an article worthy of being featured in "the encyclopedia that anyone can edit". But such concerns are unlikely to come up often enough (or at all) to be a part of the FA criteria. Ucucha 13:10, 1 February 2010 (UTC)[reply]
Disagree as per Awadewit. –Juliancolton | ^Talk 13:39, 1 February 2010 (UTC)[reply]
Oppose. I detest list-defined references in general. Given that I'm not the only one who views them in this manner, I think it inappropriate to recommend their use in FAs. Karanacs (talk) 17:26, 1 February 2010 (UTC)[reply]
Oppose; per Ucucha. But also because the proposed text is not quite clear; if citation templates are used with a short footnotes type arrangement, LDR isn't needed. It should make clear that it is applying to articles where the full citation is in the footnote. Christopher Parham (talk) 18:24, 1 February 2010 (UTC)[reply]
Oppose any attempt to make LDR preferred. Also per Finell because the appearance in the edit window should not have anything to do with FA consideration. Jim Miller ^{See me | Touch me} 19:14, 1 February 2010 (UTC)[reply]
Oppose per Ucucha and Karanacs. Why make something more complicated.? Auntieruth55 (talk) 19:59, 1 February 2010 (UTC)[reply]
Support. Regardless of the wording or what we recommend, I'd like to see something discouraging editors from adding template clutter to texts, because it makes articles very difficult to copy edit. SlimVirgin ^{TALK contribs} 10:28, 2 February 2010 (UTC)[reply]
Comment I support the use of list defined references, though this is the first I've heard of them. (Long overdue). However they're too new to require, and may not be the best solution for every article. So long as the article has consistent and properly formated citations (with or without templates), the exact method shouldn't be specified. Will Beback talk 10:53, 2 February 2010 (UTC)[reply]
Decloaking to strongly oppose this. At least there were legitimate arguments that could be made in favor of mandatory alt-text; this is a pure case of someone trying to foist their personal prejudices onto the process. – iride scent 14:21, 2 February 2010 (UTC)[reply]

So why do you have to make it personal? I'm offended. Tony (talk) 10:10, 3 February 2010 (UTC)[reply]

Not "personal" in relation to you, or Slim, or anyone else supporting; personal in the sense that – as there's no technical reason why LBR's are 'better' – making them mandatory (which in reality, "preferred option" means at FAC) is purely a case of those who support LBR insisting that we do things in a particular way. Personally, I dislike LBR intensely and find it very hard to work with – I tend to work my way through articles one section at a time, and the introduction of LBR means the references are in a separate section of their own, and thus invisible to me when editing, necessitating an irritating save text section → open reference section → edit reference section → save reference section → re-open text section routine every single time I add anything, or editing the entire article instead of a single section, with the consequent need to keep scrolling here-there-and-everywhere each time I add or amend a reference or just want to check what it says. If I were to insist that LBRs were banned (don't tempt me) I'd be just as guilty of assuming that my personal preference was automatically "right" for everyone. – iride scent 15:56, 3 February 2010 (UTC)[reply]

Oppose ~~Comment~~. Any "one size fits all" is probably wrong. IMO the following cases are not suitable for list-defined references: any ref using only once; book where only 1 page/range is cited; more journal articles that are fairly short,so that where pages/ranges are not used. --Philcha (talk) (updated to "oppose" in RFC 1, should have been copy to "oppose" here --Philcha (talk) 06:36, 19 February 2010 (UTC))[reply]
- I can say with a degree of confidence that Tony isn't supporting the use of {{r}}. WFCforLife (talk) 01:56, 4 February 2010 (UTC)[reply]
Is this RFC (2nd) still live? The discussion have spread into several topic, some technical, and some (would-be contributors) have could this confusing. If there's still a discussion, I could have a go at untangling them at bit. --Philcha (talk) 06:36, 19 February 2010 (UTC)[reply]
Support I support tony's argument. --DuKu (talk) 00:57, 4 February 2010 (UTC)[reply]
Strongest possible oppose FA is not the place to try to override and define what is the "preferred" citation style when there are more than enough legitimate ones. Basically banning any article from FA because the editor absolutely abhor LDR as one of the most hideous ideas ever implemented....ugh. Just wrong wrong wrong wrong. There is no consensus at all across Wikipedia that claims this style is even remotely better. Leave as is. -- AnmaFinotera (talk · contribs) 07:23, 4 February 2010 (UTC)[reply]
Oppose. Not only are the arguments by Awadewit, AnmaFinotera, Ealdgyth, Ucucha, Will Beback, etc. compelling, but the proposal itself labors under the incorrect assumption that not using templates (and formatting citations by hand) avoids the clutter. Featured article criteria shouldn't treat template-formatted citations differently from citations formatted by hand. Eubulides (talk) 09:14, 4 February 2010 (UTC)[reply]
Oppose. I am someone who has benefitted from using list-defined references, and participated in projects where there was agreement about gain in those cases. But I agree with concerns regarding "instruction creep", or giving any sense of it being required. It is but one tool in the editing tool-kit. Sometimes a benefit will be seen. Sometimes not. I have also contributed to articles where it isn't used. Editors should feel free to work that out for and amongst themselves. Guidance for how to think about problems is one thing. Indicating preferred solutions is another - it can shut down problem-solving. And it is probably the act of problem solving that gives most of us a buzz in contributing to Wikipedia. Over time, examples of utility and non-utility will grow, making it easier for people to work out whether to give it a shot, or if they do give it a shot, whether to continue or not. Wotnow (talk) 10:25, 7 February 2010 (UTC)[reply]

Comment. I would have less concerns about a sentence that said say, "If a citation template is used, list-defined references may help to separate reference clutter from the article and make it easier to edit. However, it is by no means a requirement". That sort of thing. It points to one way to think about things. If there's other possibly helpful ways of problem solving, I'd be appreciative of being made aware of them, and I daresay others would too. Wotnow (talk) 10:25, 7 February 2010 (UTC)[reply]

Discussion about citation templates

Eubulides, I'm surprised to see you oppose, because you've promoted the issue of ALT text as an access issue. I see this as something of an access issue too. I can't edit articles that have large numbers of citation templates in them (and they do very much add to the clutter inside texts). Perhaps other people have eyes that can separate them from the article, or perhaps other people's brains don't copy edit the way I do. All I know is that I'm effectively barred from editing articles with citation templates, and I can't imagine I'm the only one. It has meant, for example, that I've been unable to help save a couple of articles at FAR that I'd have liked to contribute to.

Is there some compromise wording we could come up with that at least makes FAC writers aware that this can be an issue, and that the templates aren't actually encouraged by the FA criteria? SlimVirgin ^{TALK contribs} 09:34, 4 February 2010 (UTC)[reply]

Hi, SlimVirgin, I sympathise as I've just had to get a stronger pair of reading glasses - and even then I have to read at the right range from the screen! But, as I said above, I think list-defined references can be good for some citations and poor for other in the same article, and at least one editor strongly dislikes list-defined references in general.

Have you tried other techniques, e.g.

CTRL plus + will zoom both text and images. It worked in Firefox and other Gecko-based browsers, Opera and even IE (although IE does not show this in its menu and users have to find the technique from external sources).
User:Cacycle/wikEd includes colour-coding of several types of text in the edit box, including citations identified by <ref> tags, and this includes citation templates within <ref> tags. wikEd works on several browsers but not IE - but I avoid IE as IMO it still has security issues for home and small business users who don't have support staff and don't lock down their IE configuration(s). --Philcha (talk) 10:13, 4 February 2010 (UTC)[reply]

Thanks, Phil. I've tried the colour-coding. The thing is that, when you're copy editing and polishing, what you're trying to produce is a smooth text, with no sharp bits, if I can put it that way. To achieve that, you need to let your eyes slide over the text again and again, in the hope that anything rough will jump out at you. Anything that clutters up the text, including colour-coding, disrupts that process. I understand that people don't want to say "please use this or that." I'm just hoping we can at least make clear that the templates aren't actively encouraged, because some editors seem to think they are. SlimVirgin ^{TALK contribs} 10:31, 4 February 2010 (UTC)[reply]
- I'm with SlimVirgin on color-coding: for me, it slows things down more often than it speeds things up. I am sad to hear that citation templates are such an obstacle; I guess they are less of a pain for me because I typically don't read the edit text much: instead, I read the formatted web page. I agree that they're harder to read in the editing window, and have proposed major (but nontrivial-to-implement, alas) improvements in that area. As for the access issue: generally speaking, readers should take priority over editors in Wikipedia. Obviously there are limits: we don't want to make things really hard on editors in exchange for relatively-minor advantages for readers. But it's not inconsistent to favor both alt text and template-formatted citations: alt text (arguably) helps readers and does not impose too much burden on editors, while citation templates (arguably) ease the editing burden overall (even if it increases it for some editors). Eubulides (talk) 11:12, 4 February 2010 (UTC)[reply]

I do agree about the need to make things easier for readers, but two points: (1) part of the point of Wikipedia is to break down the reader-editor distinction somewhat; and (2) anything that makes good writing harder is not in the readers' interests. Would you object to the criteria at least saying the templates are not encouraged? SlimVirgin ^{TALK contribs} 11:33, 4 February 2010 (UTC)[reply]

I strongly disagree with SV. If we say anything is encouraged or discouraged, the FA reviewers take that to mean it is graven in stone and handed down from Sinai. Second, I understand her point that it is difficult for her, but this does not appear to be a problem for most reviewers. And, at the end of the day, the point of this exercise is not to collect bronze stars, but to build content, and if what writers do works for them and does not cause a problem for the reader (again, the ultimate part of the exercise) then I am inclined to say, allow it. We have deliberately chosen not to standardize referencing, I would strongly object to an effort to do so through the back door.--Wehwalt (talk) 12:47, 4 February 2010 (UTC)[reply]

Note, please, that my suggestion is simply that we make clear the templates aren't encouraged, nothing more. Some editors seem to think they are, which is one of the reasons their use proliferates. I'm not suggesting we promote any particular style or format. Also, we don't judge access issues by what is a problem for most readers, editors, or reviewers, but what could reasonably be described as a problem for some of them. And again, I think it does cause a problem for the reader, because the use of lots of templates tends to lead to poor writing, or at least doesn't incline toward good writing. There may be exceptions to this, but the point is simply that it's much easier to write well without long templates buried inside sentences and paragraphs. SlimVirgin ^{TALK contribs} 12:57, 4 February 2010 (UTC)[reply]

I'll say it again: reviewers view such statements as equivalent to "you must do this". I will say that you would have to convince me that the use of templates leads to poor writing; god knows there are enough non-templated articles out there which could use some writing improvement. Show me the evidence; saying it don't make it so. I'm not certain why it poses a difficulty for you in particular, SV, there may be circumstances of which I am not aware, but we are not having reviewers draw back from articles "Begod, this article has templates. I canna review it."--Wehwalt (talk) 13:12, 4 February 2010 (UTC)[reply]

When articles have lots of templates inside the text, they tend to become lists of sentences, rather than a flowing narrative, because it's virtually impossible to achieve that flow with the constant interruptions for your brain of irrelevant text. Maybe there are people out there who can do it. I can't, and generally the evidence is that other people have problems doing it too, because the more templates an article has, the less the narrative flows, in my experience. I'm stressing that there may be exceptions, because obviously I haven't read all FAs, and I don't want to insult anyone, so please don't anyone jump on me. But speaking generally, the more templates there are, the more confusing the writing is, or the more staccato. SlimVirgin ^{TALK contribs} 13:20, 4 February 2010 (UTC)[reply]

Maybe that is true. Maybe it is not. I have no idea. Can you show us some articles (please do not use yours or mine, look for some we can both view with equanimity) to prove your point? I realize that you do not desire to offend, perhaps pick some from writers who are now inactive?--Wehwalt (talk) 13:32, 4 February 2010 (UTC)[reply]

I think I would just as soon not give examples, and in any event, isolated ones would prove nothing. It's the tendency that's at issue. I can always tell when a well-developed article contains lots of templates before I open it, because (a) it is slow to load, often to the point nowadays when I give up (the more templates an article contains, the more get added, so this is a situation that worsens over time), and (b) it is badly written in that peculiar way templates tend to produce: lists of staccato sentences, often poorly organized, because it's almost impossible to get an overview of your own text as you write it.

I think we should ask ourselves this: do we imagine that professional writers place themselves at this kind of disadvantage? Do we suppose that Philip Roth struggles to produce his perfect sentences while peering through a jungle of templates trying to find them? No. So there's no reason Wikipedians should have to. The FAC criteria encourage brilliant writing. Citation templates make brilliant writing impossible.SlimVirgin ^{TALK contribs} 13:54, 4 February 2010 (UTC)[reply]

Philip Roth doesn't need to worry about citations much, so he's not a good example. In my experience the main problem with editor-mode clutter is the need for citations, not how they're formatted.
It's true that {{citation}} and {{cite book}} etc. are way too slow. Faster alternatives exist, and more are being developed. See this performance demo. According to one (artificial) benchmark, using {{Citation}} and {{cite journal}} templates causes page generation to be about 6.6× slower than doing things by hand (10.396 secs vs. 1.584 secs). The recently developed {{vcite journal}} etc. templates are more than twice as fast, at about 2.7× slower than by-hand citations on another benchmark (4.069 vs 1.531 secs). The experimental code on that page (which does not yet work in Wikipedia) is even faster than {{vcite journal}}, at 1.2× slower than by-hand in the first benchmark (1.965 vs 1.584 secs) and 1.3× slower in the second benchmark (2.00 vs 1.531 secs). All these timings are medians of three tests, on that server. My own experience is that switching to {{vcite journal}} etc. speeds up page-load times by a factor of two in well-cited articles, and I'm hoping that the new technology will give us at least another factor of two.

Eubulides (talk) 00:43, 5 February 2010 (UTC)[reply]

(ec) Hi, SlimVirgin, 2 of wikEd's options are:

"Hide <ref>...</ref> in-text reference tags" - it represents each <ref>...</ref> tag as a small button which with one expands <ref>...</ref> a tag temporarily to seem what it is. The buttons are small and unobtrusive.
"Hide templates", which does the same to any template, including cite templates.

I use refTools to code cite templates, and have no mention of hand-formatting citations. And with "Hide templates" one can hide cite templates, while hand-formatted citations remain unhidden in their full glory. --Philcha (talk)

Wehwalt, you were saying there's no problem at FAR/FAC with templates and reviewers. Israel has just been nominated for FAR. I would like to copy edit it but can't because it's full of templates. It's also unacceptably slow to load because of them. SlimVirgin ^{TALK contribs} 16:39, 5 February 2010 (UTC)[reply]

Only takes about two seconds to load for me. --Malleus Fatuorum 17:04, 5 February 2010 (UTC)[reply]

I have a powerful computer, new in 2009, and I'm up to about 30 seconds. Retrieving diffs is even slower. SlimVirgin ^{TALK contribs} 17:07, 5 February 2010 (UTC)[reply]

I just timed it -- this is just loading the article, not a diff. 1 minute and 13 seconds. SlimVirgin ^{TALK contribs} 17:10, 5 February 2010 (UTC)[reply]

I just timed it as well, and it took 3.4 seconds to load. Mind you, I do live in an advanced, industrialised Western nation. --Malleus Fatuorum 17:18, 5 February 2010 (UTC)[reply]

(edit conflict) Slim: I edit Wikipedia on an old, 2004 IBM laptop with only 1 GB of ram (I use a more powerful computer for my real work). My load times are much faster than you describe. I normally use Firefox, but even IE 7 and 8 load Wikipedia articles much faster than you describe. You may need to look into your settings. If typical load times were that slow, lots of people would be complaining.—Finell 17:28, 5 February 2010 (UTC)[reply]

Lots of people are complaining, Finell. Eubulides has written an explanation of it somewhere, perhaps on this page; some of the templates are slower than others. Adding 292 templates to a page and expecting it not to have loading issues is not reasonable. I just tried to open it again: 22 seconds this time. SlimVirgin ^{TALK contribs} 17:33, 5 February 2010 (UTC)[reply]

I just tried it again as well; 3.2 seconds this time. There's something wrong with your setup. It's a very common problem amongst computer programmers to focus on microefficiency when really it usually doesn't matter a damn. So the load time could be improved from 3.2 to 2.8 seconds say by recoding something or other. So what? Who'd notice? Added to which most pages readers see will be coming from a cache anyway. --Malleus Fatuorum 17:40, 5 February 2010 (UTC)[reply]

I don't know what would be wrong with my set-up. I only notice these delays when there are lots of templates in an article, so it's not as though I'm having problems in general. The point is that, if one person is experiencing it, chances are high others are too. We have alt text for the small number of blind readers who might benefit, yet we allow FAs to contain hundreds of templates that are almost guaranteed to significantly slow load time for a lot of people. That makes no sense to me, especially when there are alternatives. And bear in mind that these things only get worse. An article with 200 templates today will have 250 next year if it's being edited regularly. Something has to give. SlimVirgin ^{TALK contribs} 17:48, 5 February 2010 (UTC)[reply]

Machines will be faster next year though, so no net difference. I'm puzzled though why your page load times appear on average to be more than 10 times mine. That can't have anything to do with templates. --Malleus Fatuorum 17:57, 5 February 2010 (UTC)[reply]

I just tried to load the article. The first time I did so, it took quite some time to load, but I didn't time it exactly. I then tried again with Ctrl+Shift+R and timed, but it only took four seconds or so. Not sure what caused that (it may have been either my browser or the server), but perhaps something similar happened to SV. Ucucha 18:05, 5 February 2010 (UTC)[reply]

Machines in the store being faster next year won't help people who don't own them. :) I have it with all pages that contain citation templates. I know in advance which ones do by the length of time they take to load. I've had this computer for about a year, Windows, with lots of RAM, and I had the same issue with my previous computer, a Mac. It's especially bad opening diffs. Could you open a diff on Israel and see if that's slow for you? SlimVirgin ^{TALK contribs} 18:01, 5 February 2010 (UTC)[reply]

(edit conflict) I think I see what's going on here. I just made a small edit to the article and the page took just over 30 seconds to reload. But once it was generated and cached it could be refreshed in just over 3 seconds. So it does look like there's something there that's slowing page generation down quite substantially. Luckily most readers will probably be served a cached version, but it does make editing very laborious, as you say. --Malleus Fatuorum 18:08, 5 February 2010 (UTC)[reply]

Gimmetrow and I once took a large article with templates (I don't remember which article, where or when) and removed the templates to compare load time: there was no difference. It's usually images that are slowing down load time. SV, now you know why I have a high edit count; I edit, review text, see something else, and re-edit a gazillion times. Working around citation templates is difficult, but no more so than working around short notes. Working in three different windows, with list-defined references, for me, is torture. Six of one, half a dozen of the other. We often have to emphasize that citation templates are not required, though (I also hate them, and wrote Tourette syndrome with manual citation, but I was writing that article largely alone and find it better to use citation templates on articles that are heavily edited by many other editors, as that keeps them consistent); if the FAC criteria aren't clear enough in mentioning that citation templates are optional, they should be. SandyGeorgia (Talk) 18:14, 5 February 2010 (UTC)[reply]

There's certainly a real problem with Israel's load time (or at least its page generation time) but it's nothing to do with {{citation}}. There are about 40 in that article, almost all of them in the References section. I just did a little test by deleting all of them and it didn't make any perceivable difference to the initial load time, still around 30 seconds. I suspect that your right SandyG, it's to something do with images. --Malleus Fatuorum 18:22, 5 February 2010 (UTC)[reply]

Inside the text of that article, there's citation, cite web, cite book, and cite news. SlimVirgin ^{TALK contribs} 19:21, 5 February 2010 (UTC)[reply]

(ec) It's a particular problem with articles like Israel, where the writers have to supply an abnormal number of sources, because everything is constantly questioned. The number of templates therefore grows quite quickly (this one had around 200 when promoted, and about 90 have been added since). As the number of templates increases, the article becomes slower to load, and harder to read in edit mode, so most editors wander off. The article is therefore left in the hands of the ones with strong opinions, or in the care of occasional editors. The result is that the quality of the writing inevitably deteriorates, and is impossible to fix, again because of the templates. I don't think anyone here would want or be able to copy edit that page. SlimVirgin ^{TALK contribs} 18:19, 5 February 2010 (UTC)[reply]

The problem appears to be nothing to do with templates, as I said above. Much more likely to be related to the use of images, as SandyG suggested. --Malleus Fatuorum 18:24, 5 February 2010 (UTC)[reply]

Sandy, interesting what you say about a page with and without. Perhaps if I can stand to, I'll post a version of Israel without templates in my userspace and we can compare. SlimVirgin ^{TALK contribs} 18:22, 5 February 2010 (UTC)[reply]

It might be faster to ask Gimme where/which article we did that on; he may have it somewhere in his userspace. Also, since there's something about clearing your cache for load time, I'm a techno-dummy on that sort of work :) SandyGeorgia (Talk) 18:28, 5 February 2010 (UTC)[reply]

I created a version of Israel without templates (using Special:ExpandTemplates) at User:Ucucha/Israel and without images at User:Ucucha/Israel 2. Didn't test load time yet. Ucucha 18:32, 5 February 2010 (UTC) and 18:42, 5 February 2010 (UTC)[reply]

Fascinating. Neither made much difference, both still about 30 seconds to refresh after a small change. The version without images is certainly a little quicker to load, but only a by a few seconds. --Malleus Fatuorum 19:06, 5 February 2010 (UTC)[reply]

Me too, unfortunately. I can only ever produce anecdotal evidence -- along the lines of, "I did this and this, then that happened. No, really, I promise, it did." I know we were having similar problems on Michael Jackson, where several editors were complaining about load time. I went in and removed a bunch of templates and it speeded up, though I was soon beaten back, so the experiment remains incomplete. :) SlimVirgin ^{TALK contribs} 18:34, 5 February 2010 (UTC)[reply]

There are caching issues involved that I don't fully understand; again, if Gimme knows where to find it, he did it all right, since he's not a techno-dummy like me. He might know where to locate the thread and info. SandyGeorgia (Talk) 18:38, 5 February 2010 (UTC)[reply]

For what this is worth, my PC is ten years old, but I have a very fast internet connection. Israel loaded in 5 seconds and is not cached on this PC. Graham Colm ^Talk 18:43, 5 February 2010 (UTC)[reply]

Not cached on your PC, but cached either at your ISP's proxy server and or at wikipedia, so the page will only be regenerated if it changes. Try making a small change to the article and see how long it takes to get the updated version sent back. --Malleus Fatuorum 18:54, 5 February 2010 (UTC)[reply]

Malleus, it took about 30 seconds, 6 times longer. Graham Colm ^Talk 19:11, 5 February 2010 (UTC)[reply]

I thought it would. --Malleus Fatuorum 19:18, 5 February 2010 (UTC)[reply]

Random break

This site is helpful in analyzing load times and article size and components. SandyGeorgia (Talk) 18:46, 5 February 2010 (UTC)[reply]

I'm a bit confused as to why SV is assuming it is templates on the basis of so little evidence, and more than that, is asking us to take actionwhen the evidence is so slight and she indicates that this is not an area she understands nor has she sought guidance from someone more in the know. I don't think there's anything to be done here.--Wehwalt (talk) 19:01, 5 February 2010 (UTC)[reply]

What little evidence has been gathered so far seems to indicate that the problem is nothing at all to do with templates, so I;m puzzled why they're being singled out as the root of all evil as well. --Malleus Fatuorum 19:08, 5 February 2010 (UTC)[reply]

I wouldn't blame templates either. The article is over 150KB long. I would be shocked if it loaded quickly for me. (It takes 20-30 seconds.) Geometry guy 19:21, 5 February 2010 (UTC)[reply]

In fact it's quite clear that it's the structure of the page that's the problem, nothing to do with templates. It takes 91 separate http connections to download the page, which is quite simply ridiculous. --Malleus Fatuorum 19:29, 5 February 2010 (UTC)[reply]

How can you tell it's not connected to templates? SlimVirgin ^{TALK contribs} 19:30, 5 February 2010 (UTC)[reply]

Your question makes no sense. Templates are expanded by the mediawiki software to generate html, which is what the browser is downloading. Templates aren't objects that can be connected to via http. --Malleus Fatuorum 19:35, 5 February 2010 (UTC)[reply]

(ec) That yielded the following (all for a 56 K connection):

138.92 s for Israel (current page)
139.41 s for User:Ucucha/Israel (templates expanded, but I just discovered that templates within ref tags are not actually expanded by Special:ExpandTemplates, so usefulness of this is limited)
60.87 s for User:Ucucha/Israel 2 (no images)

So it appears that images do have a very significant impact (surprise), but templates outside <ref> tags do not. Whether templates inside such tags do I have not been able to test yet. Ucucha 19:17, 5 February 2010 (UTC)[reply]

The images are indeed significant, but not the main problem with the Israel page it appears. I quote: "The total number of objects on this page is 91 which by their number will dominate web page delay. Consider reducing this to a more reasonable number. Above 20 objects per page the overhead from dealing with the actual objects description time and wait time) accounts for more than 80% of whole page latency". So while the number of images has an impact (there are 73 on this page), it's not the whole story, and it's got bugger all to do with templates. --Malleus Fatuorum 19:22, 5 February 2010 (UTC)[reply]

Did you remove all the templates, Ucucha? SlimVirgin ^{TALK contribs} 19:22, 5 February 2010 (UTC)[reply]

No. As I said, ExpandTemplates apparently does not expand templates in ref tags. However, it did expand the citation templates in the "Bibliography" section, and that did not decrease load time. Ucucha 19:25, 5 February 2010 (UTC)[reply]

Sorry, I don't understand, Ucucha. I have very little technical knowledge. All I can see is that you didn't remove all the citation templates. You would need to do that to see if it makes a difference to the load time. SlimVirgin ^{TALK contribs} 19:33, 5 February 2010 (UTC)[reply]

All the citation templates are inside ref tags. If you didn't remove those, that could explain why there was no difference. :) SlimVirgin ^{TALK contribs} 19:23, 5 February 2010 (UTC)[reply]

No, they're not. Almost all of the citation templates are in the References section, where there are no ref tags. --Malleus Fatuorum 19:26, 5 February 2010 (UTC)[reply]

Almost all the citation templates are inside the text, Malleus, and Ucucha's test page at User:Ucucha/Israel still contains templates -- cite web, etc. SlimVirgin ^{TALK contribs} 19:28, 5 February 2010 (UTC)[reply]

Are we looking at the same article? In any event, it's quite clear from the analysis that templates have no discernable effect on download times. The problem is the number of objects on the page, requiring 91 separate http connections. That's just bound to be slow. --Malleus Fatuorum 19:32, 5 February 2010 (UTC)[reply]

PS. I was specifically talking about {{citation}} templates. --Malleus Fatuorum 19:37, 5 February 2010 (UTC)[reply]

There's no point in focusing on one and then arguing it contains only 40, all in the Refs section. All that achieves, Malleus, is several pointless, confused posts. Not helpful. SlimVirgin ^{TALK contribs}

The confusion, I'm sorry to have to say, is all on your part. It is very clear what the problem is with the Israel page, and it has nothing to do with templates, no matter how many times you claim that it does. --Malleus Fatuorum 19:44, 5 February 2010 (UTC)[reply]

I'm confused as to how you can say the text contains no templates. Please open it and look; it is full of them. :) Go into edit mode and do a search for cite web, cite book, cite news, cite paper, etc. SlimVirgin ^{TALK contribs} 19:36, 5 February 2010 (UTC)[reply]

Where did I say it contains no templates? What I will say categorically though is that whether it contains templates or not the page will still take around 30 seconds to load, because the problem has nothing to do with templates. It's to do with the number of separate http connections that have to e made to download the page. And that has absolutely nothing to do with templates. --Malleus Fatuorum 19:41, 5 February 2010 (UTC)[reply]

I now fixed User:Ucucha/Israel to expand all templates. I haven't done a loading check yet, but expanding all cite templates in ref tags did add 128 Kb of text (111 pages in my text document). The source of that pages is a very bad mess when you look at it, and that mess is generated by the cite templates. Ucucha 19:43, 5 February 2010 (UTC)[reply]

Are you saying the slow load time is caused by the citation templates? Forgive my being dense.

Malleus or Ucucha, other pages that were recently used as examples of cite templates slowing load time are Heather Mills and Gaza War. Perhaps you could see whether the same issue crops up there? SlimVirgin ^{TALK contribs} 19:46, 5 February 2010 (UTC)[reply]

User:Ucucha/Israel is still 139.41 s now (figure for 56 K connection). I suspect what may be slowing the page down to some extent is the cruft for example here. This is generated by cite templates, but usually invisible because the templates are not expanded. Ucucha 19:49, 5 February 2010 (UTC)[reply]

Makes no difference for me either. Download time still about 30 seconds, as I predicted. --Malleus Fatuorum 19:52, 5 February 2010 (UTC)[reply]

I'm really confused about what's being done here. Ucucha, the argument is that citation templates (cite, cite web, cite book, etc etc) slow down load time. We therefore need to test load time with them removed entirely, not expanded. If you expand them, the extra words that we're complaining about remain in the text.

What we need to compare is (1) Israel as it is; (2) Israel with everything between ref tags removed; and then (3) Israel with the same source material restored minus citation templates. That is an experiment that will take some time to organize, because some poor soul will have to do (3) manually. SlimVirgin ^{TALK contribs} 19:58, 5 February 2010 (UTC)[reply]

The argument being put forward – that citation templates slow down load times – is patently false, or at best specious. It's quite clear from looking at the analyses of pages that load slowly and those that load more quickly that the overwhelming amount of waiting time is caused by the number of objects on the page, and hence the number of http connections needed to download them, in a quite predictable way. For instance, Israel has 90 html objects and downoads in just over 20 seconds. Heather Mills has just over half that number (48) and downloads in just over 11 seconds. If we want to improve these download times we need to be looking at the root cause, not using slow downloads as a stick to beat the completely unimplicated citation templates with. There's really no issue here; there is absolutely no evidence whatsoever to show that the citation templates have an appreciable effect on download times. Quite the reverse in fact, it's very clear that they don't. There's no need for a change to the FA criteria to address a non-existent problem that appears to have been so widely misunderstood by so many for so long. --Malleus Fatuorum 20:14, 5 February 2010 (UTC)[reply]

(2): User:Ucucha/Israel 3. Still working on (3). Ucucha 20:19, 5 February 2010 (UTC)[reply]

(ec, for Malleus) It may be clear to you that they don't slow things down, and you may be right. But it's not clear to me. And the experiment Uchuca conducted missed the point, namely that the templates add a whole bunch of unnecessary clutter. I'd appreciate if you could explain—in non-technical words of one syllable for an idiot like myself, if possible—why you are sure that "it's very clear" that these templates don't slow things down. All I know is that I have trouble loading pages with templates, and I don't have trouble loading others, including pages with lots of images. I realize that this is anecdotal, but it's a very consistent observation. SlimVirgin ^{TALK contribs} 20:24, 5 February 2010 (UTC)[reply]

You're surely not claiming that you have trouble loading every page with templates are you? What you're having trouble with is loading large pages with templates. But those large pages also tend to contain a large number of html objects, and it's those objects that are causing the problem. To download each one requires a new http connection, which involves an exchange of session information, which may require a database connection on the http server, as http is a stateless protocol. HTML objects are things like images, javascript, CSS, and so on. I've never tried it because I know what the result would be so it would be a waste of time, but you might like to experiment by making up a large page containing only templates, generating only a single html object, to prove to yourself that templates are not causing the long download times on pages like Israel. --Malleus Fatuorum 20:44, 5 February 2010 (UTC)[reply]

So, what will satisfy your concern? Would you like to see Israel without templates, or something else? Images? I still see no indication that anything needs to be changed. Should we remove one object at a time from the article? I suppose that is the truest test.--Wehwalt (talk) 20:42, 5 February 2010 (UTC)[reply]

User:Ucucha/Israel 4 has all the cruft removed from the citation templates, and is still 135.55 s. Ucucha 20:50, 5 February 2010 (UTC)[reply]

What would satisfy me, as I said above, is to see Israel asit is, and Israel with no templates in it -- no cite, no cite web, no cite book, no cite news, no cite paper, no cite anything. And then to compare the load times. The second thing that would satisfy me is to hear this explained by an expert who doesn't care one way or the other, and who isn't being aggressive, but who is just knowledgeable. Whenever I ask about this issue, I'm assured that templates do slow down load time by adding complexity to the page, causing it to take longer to parse and render. But here, I'm being told, no, it's all nonsense, I'm just imagining everything. :) SlimVirgin ^{TALK contribs} 20:52, 5 February 2010 (UTC)[reply]

Isn't that what my last test did? Also: User:Ucucha/Israel 3, with all ref tags removed completely, is 126.92 s. Doesn't make much of a difference. Ucucha 20:54, 5 February 2010 (UTC)[reply]

I was confident that Gimmetrow had already resolved this. I wish everyone engaging in this discussion could review FACs instead, so I'll have something to close :-( SandyGeorgia (Talk) 20:53, 5 February 2010 (UTC)[reply]

I'm busy with research on my next-but-one and am working on a peer review, if that is any excuse.--Wehwalt (talk) 20:54, 5 February 2010 (UTC)[reply]

Israel is slow mainly because of citation templates

The slow page-load time SlimVirgin observes in Israel is not a figment of imagination: it's real. I can easily reproduce the problem by visiting version CITE of that article (which was current as of about 15 minutes ago), clicking on the "edit this page" button, and then pressing "Show preview". The resulting page has a little comment in its HTML that says "Served by srv178 in 32.837 secs." (your timings may differ if you talk to a faster or slower server), and the entire process took about 36 seconds real time. The bottleneck is on the Wikipedia server side, not in my network or browser. If I then visit version VCITE, which is identical to version CITE except that it uses {{vcite journal}} etc. rather than {{cite journal}} etc., and do exactly the same thing, the resulting page's little HTML comment says "Served by srv227 in 12.298 secs." and the overall process takes about 15 seconds real time (again, your timings can vary quite a bit). In my experience, switching from the {{cite journal}} to the {{vcite journal}} family improves the editing experience (i.e., the real time it takes to load a page) by a factor of two. The only thing that's changed is which citation templates are used: I'm not changing images or anything else. (The test with Israel isn't quite fair, as not all the vcite flavors are implemented yet so I fall back on the cite versions, also, I approximate citation with vcite book; but it's close enough.)

This demonstration disproves the above claim "The argument being put forward – that citation templates slow down load times – is patently false, or at best specious. The citation templates are the main bottleneck in Israel, and it's their execution on the server side that is killing overall performance when editing.

Part of the confusion in the above discussion comes because some editors are visiting copies of the page that are cached on the server side. These copies are generated quickly, avoiding the bottleneck. But when you're logged in, and particularly when you're editing, the cache doesn't work.

One other thing, independent of caching: version CITE's HTML is 42% larger than version VCITE's HTML, for no particularly good reason. This difference (610 kB versus 431 kB) is not a big deal if you have a fast network connection, but most of the Internet still uses slow (non-broadband) connections, where the difference bites hard. People in (say) Africa routinely shut off images to save money (they are typically downloading over cell phones), and I expect that they routinely avoid large wikipedia pages like Israel simply because even with images turned off it's too expensive to load. This is a real and important problem with our encyclopedia. Eubulides (talk) 22:15, 5 February 2010 (UTC)[reply]

Nobody has disputed that the Israel page is slow to load if it's not been cached; the discussion centres around why, and it's self-evident from the analyses that have been carried that it's almost entirely due to the number of html objects in the page. No doubt a few microseconds could be shaved off by removing all of the templates, but so what? They're simply not the problem, and hence calls to deprecate them, or whatever change SV is proposing to the FA criteria are misguided and unjustified. your assertion that "The citation templates are the main bottleneck in Israel, and it's their execution on the server side that is killing overall performance when editing" is, I'm afraid to say, patent nonsense. Can you provide links to your test pages so that I can see for myself what it is that you're claiming? --Malleus Fatuorum 22:30, 5 February 2010 (UTC)[reply]

"No doubt a few microseconds could be shaved off" No, that's not what my benchmarks showed. Merely switching to faster citation templates shrank the time from 36 seconds down to 15 seconds, which demonstrates that the citation templates are the major bottleneck in Israel. And redoing the citations entirely by hand, as SV proposes, would shrink the time even further (I expect down to the 5–10 second range). Furthermore, my benchmarks showed a bloat of about 40% in the generated HTML, purely because of {{cite journal}} etc. as compared to better templates or doing it by hand. These improvements are not just "microseconds": they're major differences. The currently-popular citation templates have significant adverse effects on both editors and readers. I don't blame SV at all for not wanting to edit Israel: it's a terrible user interface if you have to wait 30 seconds after you press "Save page" to go on working. Eubulides (talk) 22:44, 5 February 2010 (UTC) PS. The links to the test pages are in my previous comment, but here they are again: version CITE for the current version, and version VCITE for the same page using faster citation templates. Eubulides (talk) 22:46, 5 February 2010 (UTC)[reply]

I tried these out, and Eubulides is right. {{vcite journal}} etc. is significantly faster than than {{cite journal}} etc. —mattisse (Talk) 23:35, 5 February 2010 (UTC)[reply]

Eubulides may be right that they download faster, but for the wrong reasons. I'd like to look at this myself, because I'm certain there's something else going on here, and it's got very little if anything to do with citation templates. --Malleus Fatuorum 23:39, 5 February 2010 (UTC)[reply]

I would appreciate any insight into this matter, as I understand and agree with your calls to server argument and do not fully understand the template business. —mattisse (Talk) 23:46, 5 February 2010 (UTC)[reply]

There's an overriding principle in networked systems, which is to ensure that communication doesn't overwhelm computation. Computation is cheap and fast, communication is expensive and slow. I think we need to look more deeply at why the Israel page is so slow to load, and to focus on the slow communication issue – 91 (necessarily slow) http connections for one page is simply ridiculous. If it turns out that it's the templates that are generating those 91 html objects then that's the problem that needs to be solved, but right now it's by no means obvious what the problem is. So to call for any change in the FA criteria to address a problem not yet fully understood is to say the least premature. --Malleus Fatuorum 00:28, 6 February 2010 (UTC)[reply]

The symptoms are entirely those of server-side slowdown. The slow-page-generation probelm has nothing to do with communication between browsers and Wikimedia servers. It is true that the citation templates generate significant HTML bloat, but that is of concern only to users on slow (non-broadband) connections.
It's not just Israel. It's every page that uses the standard citation templates. The more templates uses, the slower the pages are generated. Once you have more than a dozen or two citations the slowdown is noticeable. Once you have a hundred or so it gets reeeeally irritating. Once you have two hundred it is so painful to edit that many editors give up. (These numbers are approximations; it partly depends on how complicated the citations are.)
You can try the same example with just the citations, by visiting my CITEONLY sandbox, which contains (as best I can arrange it) just the citations from Israel, and my VCITEONLY sandbox, which contains the same citations using {{vcite book}} etc. In both cases, click on "edit this page" and "show preview". In my measurements, CITEONLY took about 22 seconds of server time, and VCITEONLY took 5 to 6 seconds. According to websiteoptimize.com, CITEONLY and VCITEONLY required exactly the same number (38) of HTTP requests, and the number and type of external objects was exactly the same. The only difference that websiteoptimize.com reported was that VCITEONLY generated 21,259 bytes of HTML, whereas CITEONLY generated 41,821 bytes, that is, about twice as much HTML; this is the HTML bloat alluded two bullets above.

Eubulides (talk) 06:44, 6 February 2010 (UTC)[reply]

It's a moot point anyway. The FA criteria state clearly that Featured Articles should follow the style guidelines, a.k.a. the Manual of Style. If there is consensus to change that, the criteria change automatically. If there is not, and we introduce an additional style criterion anyway, we are creating a fork of the MoS. In short, the discussion should continue, but not here. WFCforLife (talk) 02:18, 6 February 2010 (UTC)[reply]

Yes, the slowness of citation templates affects lots more than just FAs. But the slowness affects FAs more than usual, because FAs are more likely to have lots of high-quality (i.e., slower-to-format) sources. Eubulides (talk) 06:44, 6 February 2010 (UTC)[reply]

If the vcite templates are so much better, do they have a downside? otherwise, why don't we just erase the cite templates and move the vcite to the cite ones? Casliber (talk · contribs) 08:29, 6 February 2010 (UTC)[reply]

The vcite templates aren't yet complete: they cover the Big Four ({{vcite book}}, {{vcite journal}}, {{vcite news}}, {{vcite web}}) but not the other cite XXX variants. Also, these templates use Vancouver style rather than the quasi-APA style used by {{cite book}} etc. and not everybody likes this style even if it is a bit more efficient. One quick fix that could be done without worrying about style issues is to change the default in {{cite book}} etc. to not generate COiNS data and to trim away some of the relatively-useless links it generates: this would be a significant savings in HTML bloat and server CPU speed, though it still wouldn't make them as fast as the vcite family. Eubulides (talk) 08:40, 6 February 2010 (UTC)[reply]

Vancouver style citations may be common in some sciences, but not on Wikipedia, from what I've seen. In my experience, the most common citation formats on Wikipedia for full citations are the {{cite book}}, {{cite journal}}, {{cite xxxx}} family (even in manually entered citations), followed by manually entered Chicago MOS format. So, for most articles, switching to {{vcite xxxx}} templates would mean changing the existing citation style, which is contrary to the MOS guideline. But, again, this should not be an FA criterion. If the {{cite xxxx}} template family really caused slow load times throughout Wikipedia—which I very seriously doubt—"disfavoring" (i.e., prohibiting) those templates only for the tiny percentage of Wikipedia articles that are FA candidates would make only a negligible difference in the typical reader's overall experience. Plus, it seems reasonably clear that there is no consensus to adopt an FA criterion that "disfavors" citation template usage, so this is the wrong place for this discussion.—Finell 09:16, 6 February 2010 (UTC)[reply]

Citation templates do not cause "slow load times throughout Wikipedia". They cause slow load times only when the page is not already cached. They're a problem mostly for editors, not for IP readers. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

Finell, the only thing I suggested above was to add to the FA criteria that the citation templates that are causing the problems (whichever ones they are) are not encouraged as inline citations. That's it. Just "are not encouraged." One of the reasons editors add them is because they believe they're favoured. Please see that it's problematic that we're promoting articles as the project's best work when lots of people are having trouble loading and editing them, and when some people in countries with slower connections may not be able to read them at all. It's especially unfortunate when the problem can be so easily avoided. SlimVirgin ^{TALK contribs} 11:01, 6 February 2010 (UTC)[reply]

Random break 2

(outdent)As I pointed out, reviewers will take "are not encouraged" to mean "don't do it". If it is permitted by MoS, it should be permitted for FAC.--Wehwalt (talk) 12:20, 6 February 2010 (UTC)[reply]

The delegates can watch out for anyone objecting on the basis of templates alone. And anyway, we do recommend things at FAC that are not widely recommended. The advice of the MoS, for example, is not mandatory, but is routinely applied at FAC as if it were. I admit that I'm struggling here to understand why anyone would not want to say "not encouraged" of templates that are excluding people from editing, and perhaps even reading, certain articles. Just to take one example, Canada may be delisted. I wanted to copy edit that article in December when it was first posted at FAR, and couldn't because of the templates, [1] [2] and yet I'm pretty sure I could have saved it from being delisted. We really shouldn't have a class of articles that certain people are technically (or cognitively) excluded from, when we know it's happening and we could be encouraging people to avoid it. As things stand, most of the editors who add these templates to their articles don't realize they're causing problems for others. SlimVirgin ^{TALK contribs} 12:39, 6 February 2010 (UTC)[reply]

With all respect, SV, your experiences with pages you have trouble loading are—unusual. They may be unique. I believe your logic is unfounded when you say that there is a class of pages difficult to load, and we have lost FA's on account of it, unless of course you believe only you can save Canada. I suggest that you make inquiry of the people who are giving technical advice, I think at the Village Pump. Such things are much more likely to be due to the idiosyncrasies of your setup.--Wehwalt (talk) 12:45, 6 February 2010 (UTC)[reply]

Have you read any of the posts above? Israel is loading slowly for everyone. The question is what is causing it. And yes, I was the only person willing to give Canada a thorough copy edit, and I was prevented from doing it. Please don't post anything else to the effect of, "It's only you, so who cares," Wehwalt, because it's not only false; it's starting to look very rude. If you have examples or analogies or technical knowledge or insights that would be helpful, I'd love to hear them, but continually posting "not, not, not" after posts of mine is not advancing anything. SlimVirgin ^{TALK contribs} 12:52, 6 February 2010 (UTC)[reply]

These problems are definitely not "unusual" or "unique". They occur whenever the server needs to generate a page full of citation templates, which, when you're editing a page, is always. The problems are easily reproducible; just follow the steps I listed above. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

If the citations aren't in templates, but just written out, how is this any easier to copy edit? Am I missing something here? Fainites ^barleyscribs 16:20, 6 February 2010 (UTC)[reply]

Only that some editors have an irrational prejudice against citation templates that is completely unsupported by the facts. --Malleus Fatuorum 16:33, 6 February 2010 (UTC)[reply]

Somewhere on wiki there must be a knowledgable computer person who simply knows the answer to the question do cite templates make articles slower to download. Let's find them instead of debating this point.Fainites ^barleyscribs 16:44, 6 February 2010 (UTC)[reply]

The question is not a difficult one to answer, but nobody is listening. There are two issues with templates: processing time on the server and potential html bloat resulting in larger html pages. Neither of these is the reason why the Israel page is slow to load; that's quite simply down to the number of objects on the page. --Malleus Fatuorum 16:52, 6 February 2010 (UTC)[reply]

Nobody is listening because Malleus's analysis is incorrect for how editors use the page. One more data point: I just now visited Israel as a logged in editor, and because the page wasn't cached and needed to be recomputed, it took about 25 seconds to load. The HTML ended with the line "</body></html>", showing that the vast majority of this time was spent processing the page on the server. Whether this is due to the "number of objects on the page" or some other factor is irrelevant: what counts is that the page is slow to load, and (as the experiments above show) this is almost entirely due to the use of standard citation templates on the page.
"Somewhere on wiki there must be a knowledgable computer person who simply knows the answer to the question do cite templates make articles slower to download" I doubt whether there is any such person. The Mediawiki source code is messy in this area: it's obviously slow, but is contorted and will be hard to tune. I expect that the Mediawiki developers and Wikipedia maintainers would rather not think about the problem. However, there are a few signs of life: please see #Mediawiki-based solution below.

Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

Fainites, writing them manually produces extra unnecessary characters in the text. When there are a lots of them used as inline citations, it can make the text difficult to edit, because it's hard to see where one sentence ends and another begins. Examples below:

A long book reference written manually: :<ref>Amis, Martin. ''The Pregnant Widow''. Jonathan Cape, 2010, p. 1.</ref>
A short book reference written manually:<ref>Amis 2010, p. 1.</ref>
The same words as (1) with a citation template: <ref>{{cite book|last=Amis|first=Martin|title=The Pregnant Widow|publisher=Jonathan Cape|date=2010|pages=1}}</ref>
The usual thing produced by a citation template, where editors are inclined to fill in lots of parameters simply because they're there (and it's sometimes considerably worse than this example): <ref>{{cite book|last=Amis|first=Martin|title=The Pregnant Widow|publisher=Jonathan Cape|location=London|date=February 4, 2010|edition=First|pages=1|isbn=0224076124}}</ref>

Multiply the extra characters by the number of templates (usually 100 plus in a decent article; 290 in Israel, the article we've been discussing), and there's a significant problem with reading the text in edit mode, and with loading time. See User:SlimVirgin/templates for some more examples. SlimVirgin ^{TALK contribs} 16:53, 6 February 2010 (UTC)[reply]

A long time ago I compared the load times of some articles with cite templates to the same articles with the citations written without any templates. The difference in load times was not significant, in my opinion. I just did the same for the page in question here. Server load times were .054 to .158s for both pages, but without a lot of sampling I couldn't say there was any difference. Actual load and render time after a cleared cache on the particular computer I was using at the time: 4.5s to 5.0s for either page. The no-cite version may perhaps render .3s faster. Maybe. If the no-cite version is actually faster, some of that difference would due to the no-cite version having about 8% fewer characters of wiki text (or about 4% fewer characters of html); the no-cite version also has somewhat fewer html tags, which might provide a few microseconds advantage, too. If anyone wants to check these tests, see User:Gimmetrow/test and User:Gimmetrow/test2. diff. I tried to replace the cite templates with equivalent text, rather than use the short refs mentioned by SV. Gimmetrow 17:01, 6 February 2010 (UTC)[reply]

This analysis is using cached times, for IP readers. But that's not the problem. The problem is when editors are logged in: often the pages aren't cached for them, if they have preferences set or whatever, so they need to be recomputed, even if they're just reading the page. And when they're editing the page, of course the page needs to be recomputed. Please redo the analysis assuming that the pages have not been computed and are not cached. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

Gimmetrow, what's your response to the information Eubulides provided about this, at the top of this section: Wikipedia talk:Featured article candidates#Israel is slow mainly because of citation templates? SlimVirgin ^{TALK contribs} 17:11, 6 February 2010 (UTC)[reply]

My first thought is that "preview" is not the appropriate test. I wouldn't be surprised that previewing text with a lot of templates takes longer, but preview text shouldn't be server-cached. Once it is server-cached it should be just like any other page to readers. Gimmetrow 17:17, 6 February 2010 (UTC)[reply]

I don't think he's just talking about preview. But as an additional point, if you agree that previewing text with templates takes longer -- and editors have to use preview more than usual when editing with lots of templates, because it's hard to see where one sentence ends and another begins, so you're constantly having to check -- that would suggest you agree that editing with templates slow things down. Or have I misunderstood? SlimVirgin ^{TALK contribs} 17:34, 6 February 2010 (UTC)[reply]

I don't see much significant difference in load-time to a reader, but I wouldn't be surprised if preview takes longer to load with templates. Are you making an argument just for editors? Wouldn't any slowdown be mitigated by editing the page in small sections, as other factors tend to encourage anyway? Gimmetrow 17:45, 6 February 2010 (UTC)[reply]

Of course editing the page in sections will improve page-generation times, since you needn't wait for the whole page to be recomputed. However, one cannot always do that (some changes require access to the whole page, or you're editing the citation itself); besides, when one hits the "Save page" button, one has to wait for the 30 seconds or more to get the page regenerated, and this wait is seriously impeding editing. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

There are several different issues, G. First, with all the extra characters in the text from the templates, it becomes impossible to copy edit for flow (regardless of load time), because you can't see the text clearly. Secondly, you have to use preview constantly to see where the sentences start and end. Third, load time, both section editing and whole-page editing, is very slow with templates. The issue we're trying to establish here is what is causing the slowdown. Eubulides is saying it's the templates. Others are saying it's not. But the two arguments aren't addressing each other's issues, so those of us with no technical knowledge can't follow which seems to make most sense. All I know is that I always have trouble loading pages with lots of templates—Israel is taking me 22 seconds to over a minute to load—but I don't recall having trouble loading anything else, including pages with lots of images. And I have a new and powerful computer and a good connection. SlimVirgin ^{TALK contribs} 17:54, 6 February 2010 (UTC)[reply]

File:Template issue example 1.JPG

WikiEd with templates visible; I defy anyone to copy edit this well.

File:Example template issues 2.JPG

With templates hidden, an improvement but still not ideal

Following Gimmetrow's suggestion, I tried out Israel in sections to edit and preview, and found the rendering instantaneous, although the entire page does take a long time to load the first time (before it is cached), much like Major depressive disorder, for example, which can take even longer than Israel. Also, if you use WikiEd, you can hide the references so that you do not see them in the previewed text—just a little symbol showing where they are. Have you tried WikiEd? You can shut off any of its functions that you do not like. This image shows an example of WikiEd.

—mattisse (Talk) 18:24, 6 February 2010 (UTC)[reply]

Thanks, Mattisse. I've tried the wikiEd thing (see right), but most people don't have that turned on, so we can't rely on it. Secondly, I find it difficult to edit with, and it has made my browser stop responding several times. Per ACCESS, we shouldn't really be adding things to articles that require people to use this kind of tool just to be able to edit them. SlimVirgin ^{TALK contribs} 18:48, 6 February 2010 (UTC)[reply]

Rendering slowness / templates

Here is what I know about it technically:

Templates by themselves do not make loading an article slower. The templates are expanded by the servers when the page is saved, not when it is viewed.
The normal citation templates generate COinS metadata which makes the html sent to viewers quite a lot longer and more complex. The vcite templates by Eubulides do not, and I suspect this is why they display faster. I do not know what COinS is really useful for.
If you want to make Israel and other country articles render faster, I suggest taking a look at the navigation boxes hidden under "Articles Related to Israel" at the bottom. That is a lot of html to transmit and render for every page view, and how useful is it really?

--Apoc2400 (talk) 19:08, 6 February 2010 (UTC)[reply]

Thanks, Apoc, that is helpful. Regarding the COinS metadata, is there any way to find out how much load time they would add if, as in the case of Israel, nearly 300 templates that generated that metadata had been added to the article? SlimVirgin ^{TALK contribs} 19:34, 6 February 2010 (UTC)[reply]

Apoc's analysis is incorrect for editors. It is correct for IP readers, but logged-in editors need to have the page recomputed, and that is why pages like Israel routinely take 30 seconds or more to be viewed. This has nothing to do with the nav boxes; as described above, I can make Israel reload much faster simply by redoing the citations so as not to use the standard citation templates. The COiNS metadata is a part of the slow-load-time problem, but it's not most of it. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

I don't believe that you can, or have done. You've simply got a misunderstood fixation about citation templates. --Malleus Fatuorum 23:01, 6 February 2010 (UTC)[reply]

(This started off as a targeted response then overwhelmed me, so I'm just dumping it here as a general comment)

You guys are talking about several different things within the same breaths:

One issue is how long it takes for a reader to see an article. This depends mostly on the size of the page that is actually rendered in HTML after the wiki-software gets through with it, which page goes to a "squid" cache so it can be served up quickly for the next reader. In this scenario, the number of "objects" your browser downloads is not especially material though there is some extra transmission overhead for each object. The COiNS HTML overhead will obviously affect transmission time, if you "view source" of pages, you can figure out how much HTML data has been squeezed into the intertube.

Recently I believe they've moved the wikibits images to a dedicated server, which is incredibly annoying ATM because it seems that server doesn't close connections properly and it takes forever for the page to actually finish. Keep in mind though that your own browser also caches objects. Theoretically the wiki-globe logo only ever got downloaded once to your system. The time reported within the HTML source that Eubulides (I think) mentions as "server time" - that I think is the best indication of time-on-server, as opposed to time-to-average-reader once the page is squid-cached.

If you the editor have some different setting from the default for things like thumbnail size, skin &c., you will bust the squid cache and the HTML has to be rendered again from the parser cache (which has the representation of the article itself sans menu wrappers). This will take some time, although the squids will cache the pageview for the next reader with your exact same preferences.

Now if you're actually trying to edit the page, more factors come into play. When you edit (or purge) a page, you invalidate every single cache (in preview mode you're just pretending but the same stuff happens, just for you), it all starts again at the parser:
- Sheer size of the wiki-text itself will affect load time, especially for dialup connections.
- Template expansion. Anytime you put something within {{squigglies}}, you are telling the MW software to swap in the contents of the template page with appropriate parameters AND THEN SCAN THE PAGE AGAIN to see if one of the new parameters contains a template, which then gets expanded the same way. This is a recursive (re-entrant) function from my reading of the software and works at the page level - so the depth of template calls can become significant. Within the HTML source of a page you should be able to find something like "NewPP pre-expand size" and post-expand size (and the related "expensive parser functions"), these give a partial clue of how onerous use of templates is. This has nothing at all to do with your connection speed.
- Database hits. A basic limiting factor in rendering a wikipage from scratch is how many times an actual disk-page must be retrieved. From that standpoint, when you use the {{convert}} template the first time, the servers have to get the actual copy of that file from the hard drive, then go through the recursive thing I talked about above. In the case of {{convert}} I know for certain sure that the server will have to also fetch at least one sub-page of the main template. But DB / parser-servers cache pages too, if my preview task happens to get assigned to a server where you've just referenced the same objects in the database, we'll likely be able to get that good 'ol edit conflict. Again, not connected to how fast your connection or computer is.

Bottom line: for lag time from the editor standpoint, I think this is an interaction between page size, number of templates (which seem to force the entire page to be re-parsed), and depth of template calls (which force a re-parse, only recursively). I think I saw Z-man commenting somewhere, he would be well-armed to blow me out of the water here. But that's my impression of how the whole thing works... Franamax (talk) 02:28, 7 February 2010 (UTC)[reply]

So, will SV's suggestions make a noticeable difference if implemented? And if so, how much? I will not pretend to have understood everything you said, my brain blew several fuses while reading it and I need to go find replacements, excuse me.--Wehwalt (talk) 02:32, 7 February 2010 (UTC)[reply]

[no ec, but intended to post below MF 23:01] I've found this discussion interesting because of the variety of factors involved and how people tend to talk past each other. My precis: Eubulides has presented simple-to-verify evidence by way of his or her links that one citation template leads to a slower page-loading time than the other, when the page has to be "re-rendered" by the server. A page has to be "re-rendered" when it is saved, to some extent when it is previewed (editing by section ameloriates this), and from my experience with Eubulides' links, when an old diff is accessed (possibly, again, only when as a logged-on user). A reduction in HTML page size using new templates has also been demonstrated, and while this is of course beneficial, differences in HTML size have little to do with the slow-editing problem that started this discussion—the issue is server-side. Yes, we are talking about an issue that affects editors, not readers (smaller HTML pages would actually benefit the average IP reader more). SlimVirgin makes a case that long pages with many citations are hard to edit; looking back, I agree, and now I understand why they were. There is evidence that changing how a citation template is "programmed" significantly improves an editor's wait times when previewing and saving heavily cited pages. In ignoring the evidence presented while speaking of intuition and others' "fixations", Malleus appears as the one with the fixation.

All things being equal, and without regard to what citation templates are used, why not improve the citation template "programming" to improve page rendering? Eubulides has already demonstrated that it can be done. I don't like the Vancouver system—finding it pointlessly terse and unformatted when space is not a consideration here—but whatever makes the vcite template faster to render—can it not be ported to the citation templates already in use? Whether that means removing parameters that add to programmatic and editorial complexity, removing "coins" metadata, or other technical differences between "cite" and "vcite" that haven't been determined yet, or mentioned here, it seems like a way forward. (I certainly carry a bias that "debulking" Wikipedia is never a bad thing—procedurally, technically, etc. Almost everyone has a pet project that bulks Wikipedia up in areas where we don't need it, and not in content, where we need it. That's the slow death of any organization.) Outrigger (talk) 02:37, 7 February 2010 (UTC)[reply]

(fuses replaced)So can we do anything about this here or should seek to start a discussion at meta?--Wehwalt (talk) 02:45, 7 February 2010 (UTC)[reply]

That precis is a good one; thanks. The standard citation templates could be sped up quite a bit as follows:

Drop COiNS. Hardly anybody uses it, and the performance penalty hurts everybody (particularly editors).
Drop |first1=, |last1=, etc. Just use |author=.
Drop the unnecessary and repetitive wikilinks to ISBN, DOI, etc.

The result still wouldn't be as fast as {{vcite book}} etc., but it'd be a good sight faster than what's there now, and the resulting HTML would be within shouting distance of the size generated by {{vcite book}}. Eubulides (talk) 02:49, 7 February 2010 (UTC)[reply]

I don't particularly care about other article editors, so long as you see slower performance than me. :) Seriously though, I'd be more concerned with excluding dialup editors than the general problem of an article that totally pooches out no matter who tries to edit it. Eub, you've done some tests that seem to demonstrate the case empirically and quite well. Using {{vcite}} rather than {{cite}} seems to work well. However you also propose a concomitant loss of functionality, and that is what should be examined. If microformats should be discarded, shouldn't that be a standalone proposal? I recall jumping on someone once for breaking microformat data, so I'd be interested in the outcome. Dropping extra parameters saves you time on the first pass through the loop, I think you or I could verifiably demonstrate a savings of several milliseconds but not likely more. Repetitive wikilinks will take more time "on-the-wire", lets say an extra 200 bytes per link, times 200 links - that's about ten seconds on dialup so it becomes significant. Who suggested going to ask the serious pros above, might be a good idea? I do believe though that it is not so much about using a template repeatedly as it is how complex the template is in terms of its expansion (remember that all a template is, is a piece of text that I keep reading over and over again, writing in more stuff, until it doesn't have squigglies anymore, then I parse it as wiki-text). I suppose now I have to actually do that analysis rather than yapping about it. :( Performance gains generally come from either faster computers and networks (we wish); divesting functionality (always easy, except for those who relied on it); or innovations in the code which preserve functions (in which case, pls demonstrate). I'll note that while I'm interested in this discussion, I don't want to get in the way of you guys pursuing your solution, so please only include me if I'm actually helping. I don't want to get in the way. Franamax (talk) 03:45, 7 February 2010 (UTC)[reply]

Cleaning up and simplifying exisiting templates sounds good to me. Never saw the point of first/last or DOI/ISBN links.Fainites ^barleyscribs 15:29, 7 February 2010 (UTC)[reply]

Subpage ?

Gettin' frustrated by my watchlist :) The proposal has not gained consensus, and while the discussion is important, it's technically over the heads of many of us and is really not a FAC issue. Would it be possible to sub-page this thread and link back to it here? SandyGeorgia (Talk) 18:43, 6 February 2010 (UTC)[reply]

I don't mind where it goes, Sandy, but I feel it's quite an important FAC issue. We're producing FACs that aren't as well-written as they could be, yet which can't be copy edited easily, and in some cases can't be loaded easily, or at all by people in certain parts of the world. And it's happening only because editors at FAC mistakenly believe citation templates to be de rigueur. Some are even asked by reviewers to add the templates to perfectly well-formed references, and they do it because they think they have to, thereby rendering the article harder to access in future by others. SlimVirgin ^{TALK contribs} 18:54, 6 February 2010 (UTC)[reply]

I (and other FAC regulars) try to catch and combat that faulty perception whenever it occurs. If we need to beef up the instructions, then we should do so, but linking to a guideline page ought to cover it (no guideline requires the use of citation templates). My concern is that sorting this loadtime issue has become very technical, not related to FAC. SandyGeorgia (Talk) 18:58, 6 February 2010 (UTC)[reply]

... and it's not related to the use of citation templates. It's very vexing that SV still claims that it is, despite all the evidence to the contrary and absolutely non in support of her prejudice. --Malleus Fatuorum 19:03, 6 February 2010 (UTC)[reply]

There is zero "evidence to the contrary" for Israel. All the evidence shown above, for Israel, confirms SV's claims that it's the citation templates that are making the pages to load. For example, version CITE (with the current templates) took about 36 seconds to preview, and version VCITE (with the {{vcite book}} family) took about 15 seconds. The standard citation templates are clearly causing the server's poor performance with Israel. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

My vexation is that we happened to have five different RFCs appear simultaneously, and this one isn't related to FAC, so we need to clear the page so we can focus on other things. If this had been the only issue to come up on WT:FAC, it wouldn't be quite so vexing to see my watchlist. SandyGeorgia (Talk) 19:07, 6 February 2010 (UTC)[reply]

Malleus, what do you call Eubulides information if not support? I can't understand why you're being so rude about this. Disagree by all means, but why so aggressive? SlimVirgin ^{TALK contribs} 19:30, 6 February 2010 (UTC)[reply]

You surprise me, SV, by referring to Malleus, too, as "rude". Oh for two. Perhaps we need authoritative opinions on this more than name calling? A post at the village pump might be helpful. Note that I do not specifically disagree with you, I am waiting to be convinced one way or the other. I note that anecdotal evidence based on one user's experience that no one else has chimed into to say that they have experienced similar problems are not convincing. I would like to see things authoritatively stated before we glom on an addition to the FA instructions that may not be helpful. Those who have posted about technically seem to disagree considerably, and we are left with the original point that you have problems when it comes to loading for Israel. Canada too, I gather, eh? We need authoritative answers. SV, you are the one who are reporting this difficulty, can you make appropriate inquiries and report back?--Wehwalt (talk) 22:32, 6 February 2010 (UTC)[reply]

"no one else has chimed in" This is incorrect. I have experienced similar problems for quite some time. I just now visited Canada and it took about 23 seconds to load (I didn't try to edit anything: I merely visited the page as a logged-in user). The last line of HTML said "</body></html>", which means the vast majority of the problem is on the server side, and have nothing to do with my browser or network connection. Please don't dismiss these problems as not being "authoritative". They are real, and are often complained about, but for some reason complaints like these are often dismissed. Eubulides (talk) 22:53, 6 February 2010 (UTC)[reply]

Perhaps we misunderstand. I do not question that some pages are slow to load. I just loaded Canada, it took 20 seconds. My concern is with the contention that this is due to citation templates, and that making a change to another version will materially change wait time.--Wehwalt (talk) 23:12, 6 February 2010 (UTC)[reply]

The contention has been verified by taking one slow-to-load page (namely Israel), replacing {{cite book}} etc. with {{vcite book}} etc., and noting that this decreaded page-load times by more than a factor of two; see version CITE vs. version VCITE. In the benchmark noted in Wikipedia:Centralized discussion/Wikipedia Citation Style #Demo of specific proposal are comments about a benchmark in which a page with standard citation templates takes 6.6× longer to generate than a page with citations formatted by hand (and 2.4× longer than a page with {{vcite book}} etc.). There really is no doubt that most of the slowness is due to the citation templates. If you're not yet convinced, I can easily do another test based on Canada. Eubulides (talk) 23:36, 6 February 2010 (UTC)[reply]

I will take your word for it, as a competent authority. Which is what I had really been asking for.--Wehwalt (talk) 02:57, 7 February 2010 (UTC)[reply]

Mediawiki-based solution

rannṗáirtí anaiṫnid is testing, on another web site, a modification to Mediawiki that would let people use a syntax like citation templates, but without their huge performance penalties. I hope that something like this can be folded into Mediawiki, because the current performance of the citation templates is pretty bad. For more on this please see Wikipedia:Centralized discussion/Wikipedia Citation Style #Demo of specific proposal. Eubulides (talk) 22:20, 6 February 2010 (UTC)[reply]

For myself, I've had quite enough of the ignorant stupidity so evident in this discussion, so I'm glad it's been hived off to a dark corner that I can now ignore. --Malleus Fatuorum 23:08, 6 February 2010 (UTC)[reply]

Malleus, be nice (for me? :) It may be hived off, but we're still following :) SandyGeorgia (Talk) 23:26, 6 February 2010 (UTC)[reply]

Be that as it may. I read this page to see that SV made a proposal, and it was overwhelmingly rejected by community members. What are we discussing here? The same thing that has been rejected? Something else? It is customary when a propsal is rejected to allow some sort of decent interval before bringing it back. —Preceding unsigned comment added by Wehwalt (talk • contribs) 17:38, February 6, 2010

Tend to agree here. We all get that SV hates citation templates, but that seems like a personal issue rather than anything to do with FA. The proposal itself has long been rejected. -- AnmaFinotera (talk · contribs) 23:40, 6 February 2010 (UTC)[reply]

Wehwalt, I have not made a proposal. I opened a discussion about whether to add some kind of "not encouraged" language to the criteria. Once we determine (if we determine) what is causing what, I may make a proposal. Two of you seem very hostile to it even being discussed, for reasons I can't fathom. That's fine, but please allow others to talk about it without the constant barbs. SlimVirgin ^{TALK contribs} 01:11, 7 February 2010 (UTC)[reply]

Doesn't the proposal which was !voted upon at the top of the page entirely dispose of the matter? And please do not attribute emotions or positions to other people, speak only for yourself.--Wehwalt (talk) 01:22, 7 February 2010 (UTC)[reply]

You are not reading the posts you are commenting on. :( Tony proposed recommending another kind of template, if people want to use templates for inline citations. That was rejected. I then opened a discussion about something else, about whether to add some words of non-encouragement. So now we are trying to discuss what is causing the slowdown. Please allow us to do that! SlimVirgin ^{TALK contribs} 01:26, 7 February 2010 (UTC)[reply]

Perhaps if you would take up my suggestions as to how to prove or disprove your case, such as seeking assistance outside of this hothouse, you might be able to proceed more efficiently.--Wehwalt (talk) 02:10, 7 February 2010 (UTC)[reply]

Resolved?

Any objection to marking this page resolved? Discussion of the technical fixes won't take place here and I am not certain what can be accomplished here. However, if others feel there is a useful purpose to more discussion on this page, I will certainly withdraw the suggestion.--Wehwalt (talk) 03:04, 7 February 2010 (UTC)[reply]

If something is determined, can someone let us know somewhere? I can't follow most of what's being discussed. SandyGeorgia (Talk) 03:49, 7 February 2010 (UTC)[reply]

It isn't resolved. I would appreciate it if Wehwalt would stop trying to prevent us from discussing it. He is welcome to take it off his watchlist. SlimVirgin ^{TALK contribs} 11:11, 7 February 2010 (UTC)[reply]

Please do not tell people whether they should or should not be a part of a discussion. And kindly read entire posts, in which I said if there was objection, I would not mark it resolved.--Wehwalt (talk) 12:39, 7 February 2010 (UTC)[reply]

Why this matters

The reason this matters, and the reason having this page to discuss it is a good idea, is as follows.

We are currently creating a class of featured articles that, in future, no one will be able to copy edit because they are full of these templates. Some of them contained the templates when they arrived at FAC. Others didn't, but the nominees were told to add them by reviewers. The reviewers had no right to require them, but nominees don't always realize that. Some reviewers may believe they're required too.

The result is articles like Israel and Canada, both at FAR with writing problems, neither of them correctable. They can't be copy edited for two reasons. First, the text is impossible to read in edit mode because of the clutter of the templates. Secondly, the large number of templates causes the articles to load very slowly, particularly in preview mode. See explanations above from Eubulides, Franamax, and Outrigger, or here, but the bottom line seems to be, quoting Franamax:

Anytime you put something within {{squigglies}}, you are telling the MW software to swap in the contents of the template page with appropriate parameters AND THEN SCAN THE PAGE AGAIN to see if one of the new parameters contains a template, which then gets expanded the same way.

So the templates cause a slowdown, sometimes a significant one with loading times of 30-60 seconds. Yet to edit articles containing lots of templates, you have to use preview more than usual because it's otherwise very difficult to see where one sentence ends and another begins. It is slightly easier to edit them with WikiEd, but that introduces other problems (smaller edit window, lots of colours in the text, browser crashes).

What happens with FAs like Israel is that they are continually expanded, in part because it's the nature of the beast, but also in part because there's nothing else that can be done with the article. Because of the templates, it can't be tightened. It can't be copy edited for flow. It can't be rewritten, except by starting from scratch. The new additions can't be added in a way that will make the text around them flow well. The writing therefore inevitably deteriorates over time. But when these articles arrive at FAR, if the writing issues are serious, they can't be fixed—unless someone is willing to go in and replace all the templates.

The situation we have at the moment is therefore not sustainable. We are creating a mess for future generations of editors to deal with. That is why we need to discuss it now, and try to find a solution. SlimVirgin ^{TALK contribs} 11:27, 7 February 2010 (UTC)[reply]

More discussion: So perhaps we should charge a five cent deposit on templates? SV, just by talking about it briefly here the more technically adept editors here came up with a solution that took most of the wind out of the sails of your complaint. Presumably with a little effort they could come up with a way of making cite just as efficient as vcite. Or come up with another way of doing it. That is the proper way of doing things. And if your nightmare scenario comes true and the rivers run white with discarded plastic bags while birds are choked by those plastic things that hold six packs together, I'm a-betting that someone can write a very little bot that converts all the templates you want. SV, not one of the technically adept editors who were kind enough to favor us with their expertise said, "zOMG, SV is right, vcite is the way to go or the heavens will fall.".

Note also one thing: It is fallacious to say "oh, nothing can be done except expand such articles". That's contrary to fact. I may not watch any FA's quite as prominent as Israel (that is a "beast", as you term it, that is a bit too contentious for me), but I am zealous about keeping unneeded material out of my FA's. And if your nightmare scenario came true, and the rivers began to run white with bags from Safeway, I am sure that editors who care about that article would make sure that the problem was abated. Or else other editors would insist that it return to earlier limits, or borders as a condition for remaining a FA.

You may not like what I have to say, but I see no reason to enact a proposal for the (rather doubtful!) benefit of er, future generations. The interpretation I have from the technical conversations is that our answer lies in more efficient templates: we need not slay that beast.--Wehwalt (talk) 12:58, 7 February 2010 (UTC)[reply]

It has been a feature of all your posts here that you don't read what others say, and then you set up straw man arguments. I have not made a proposal, and I have not suggested that vcite is better: I don't even know what it is. My aim is simply to open up a discussion about the implications of using citation templates in FACs. That's all I'm going to say, because your aim seems to be to close down this discussion, or to make it impossible for some reason. SlimVirgin ^{TALK contribs} 13:01, 7 February 2010 (UTC)[reply]

I have strongly suggested, SV, that you avoid attributing motives to others, and avoid characterizing them. Please stick to the points. This discussion could lead to a considerable change in how we do FA's, I certainly am entitled to weigh in. If my concerns are germane, please respond to them; if they are not, you need not respond. Surely others are as capable as you of disregarding non-germane arguments? I will summarize my point: The answers here do not seem to lie in FA practice, but rather in technological fine-tuning. I proposed marking this resolved in good faith, because the technically adept appeared to think they could go a long way toward addressing concerns.--Wehwalt (talk) 13:10, 7 February 2010 (UTC)[reply]

"the more technically adept editors here came up with a solution that took most of the wind out of the sails of your complaint" Which solution was that? I didn't see it. Eubulides (talk) 06:39, 8 February 2010 (UTC)[reply]

Previous discussion on Wikitech-l

There was an exchange in January 2009 on the Wikitech mailing list about how citation templates cause a slowdown; see here. For a summary of the technical points on this page, see Wikipedia talk:Featured article candidates/Citation templates (technical). SlimVirgin ^{TALK contribs} 16:22, 7 February 2010 (UTC)[reply]

I can't comment on the load times, but the problems you describe as making editing difficult or impossible for future generations aren't going to go away if templates were removed:

We'll never divest ourselves of using end/footnotes to include full references, citation or not. We will always be using the <ref> code to implant refs into the article bodies, whether a spelled out citation or template based. That will always impact editing, though we can minimize it if we were able to encourage list-defined references (though there seems to be strong resistance against that.)
Moving away from citation templates will not change the complexity of looking through the text of an article to find things. To me, the templates are better because I can recognize the gobbletgook that are part of them (the "|" and "=" and curly braces) and my mind already drops them out for me, though I agree this doesn't help when working through for overall prose editing.
The basic fact is: as long as we're forcing ourselves to support an ASCII text editor (even with the improvements with Wiked) it will always be difficult to just simply look at the wikitext and understand what's going on, cite templates or not. If we were to gain a more full-fledges WYSIWYG editor that would act like MS Word in regards to references, every point you make above (ignoring the technical problems) would be satisfied. Can this happen in the future? Maybe, I believe there's a lot better support in the upcoming HTML 5 for handling better text editors, but we're a far way from supporting that on WP.

In reality, in-line use of references (template or not) really get in the way of overall copyediting and for overall article restructuring. But, I wouldn't be using WP's built in editor for that, I'd turn to AWB, to Word, or something else. It would be really nice if there was a way to export and import articles into Word or comparable formats for these more indepth editing aspects, but then you have to quickly realize how that would likely screw up article history. Wikipedia's basic editing structure is at odds between having easy-to-read Wikitext and its technology, and that's an area I can't see us resolving any time soon. The best solution, from strictly making it easier to follow text, is to enforce LDR so that reference use is clear but can be glossed over. That's regardless if it is spelled-out citations or templates. —Preceding unsigned comment added by Masem (talk • contribs) 21:51, 8 February 2010 (UTC)[reply]

I see what you're saying and agree to some extent. But it's clearly easier to read an article that has fewer words in the inline citations. So <ref>Smith 2010, p. 1</ref> is better than <ref>Smith, John. Name of Book. Cambridge University Press, 2010, p. 1.</ref> And both are better than <ref>{{cite book|last=Smith|first=John|title=Name of Book|publisher=Cambridge University Press|date=2010|pages=1}}</ref> (multiplied by 200 or more). That's my only argument here, that templates = maximum clutter for editors, plus the loading time issue. To argue "but we'll always have clutter, there's no point in fighting it" feels to me like saying "we'll always have crime, so let's not lock our back doors. SlimVirgin ^{TALK contribs} 22:09, 7 February 2010 (UTC)[reply]

I don't disagree that the short version above is better than inlines. The short version is nearly equivalent to the LDR version: <ref>Smith 2010, p.1</ref> compared to <ref name="Smith 2010 p1"/>, and the LDR version is now neutral to if the ref is a template or spelled out citation, which is why I think the better effort is to help get the word out about why LDR is better that achieves the same goals you have above with easing references. (I will note that the short ref format you use is great if a work is used multiple times, but for many articles that touch one ref and move on, you'll still end up spelling things out a lot, and not always is an author known, so that makes things a bit more difficult - It's great for the works that are based on academic-like sources but not more contemporary ones). --MASEM (t) 22:58, 7 February 2010 (UTC)[reply]

Yes, I agree, and bear in mind with manual refs you only have to type out <ref name=Smith>Smith 2010, p. 1</ref> once, and thereafter it's the same as LDR: <ref name=Smith> So they are very close in achieving minimum in-text clutter. The difficulty is that lots of editors believe citation templates are encouraged or even mandatory, and it's very difficult getting the word out that they're not. My idea here is to suggest that something be added to the featured article criteria that asks editors to avoid in-text clutter with templates if possible. That would get the word out from the top down, as it were. SlimVirgin ^{TALK contribs} 23:20, 7 February 2010 (UTC)[reply]

Well, given the above strong opposition to "prefer LDR", I don't think we can force that. Mind you, no one is stopping you from using it in your own articles, but I think until there's strong impetus to use it, we'll have to treat it like US/UK spelling and US/EU date formats (that is, first author set prefs but can be changes by page consensus) --MASEM (t) 00:04, 8 February 2010 (UTC)[reply]

Parsing templates takes a while - especially the nested templates, because they have to be expanded recursively to resolve. It appears the problem is that editors with preferences that don't match a cached version of the page, which therefore requires processing. On the wikitech list, Alex suggested [3] taking the cite templates away from the template parser, and implementing them through html tags. With this suggestion, refs would have a syntax something like:

<ref author="Smith" title="Book of Smiths" type="book">Pages 10-12</ref>

where the "book" type would be implemented in the ref extension. It was claimed that implementing cites this way would fix most of the speed issues. Assuming this would address speed issues, would such a syntax still chunk up the text too much for prose editing? Gimmetrow 02:11, 8 February 2010 (UTC)[reply]

It would be a lot better than the current situation, GT. Basically, the fewer words inside the text that aren't part of the text, the better. The problem with the templates is twofold: first, the number of extra characters contained in the templates themselves (the squiggles, the "cite web" the = signs etc). And then also the number of parameters they offer, because lots of editors will fill in whatever parameters they can. So you end up with long book citations giving the editor's name and the title of the series, and so on, all unnecessary. So the smallest possible number of characters plus minimalism when it comes to the parameters that are offered would be ideal. SlimVirgin ^{TALK contribs} 02:18, 8 February 2010 (UTC)[reply]

Well, filling in extra parameters are not always wrong. If you're citing one chapter of a multi-author book, adding the editor is usually required, for example. And filling in extra parameters will not be solved by moving the templates to a wikicode markup - just as we have example code that lists all parameters for the templates, they'll be the same for the markup. But, as long as named references still work (eg, you only have to define the details once), and still LDR compatible, this solves the other technical problem of speed. We're still left with what's best called the "social" issues of editors blinding using the templates/markup code without knowledge of what best fields to use, but its a few steps better. --MASEM (t) 02:25, 8 February 2010 (UTC)[reply]

Yes, exactly. We need both a technical and an editorial/social solution. The extra parameters don't matter (except in terms of speed) when they're in a References section. But having chapter numbers, series editors, ISBN numbers etc inside a text is very difficult to work around, especially when there are dozens or even hundreds of them. But the editors who do this aren't aware that they're causing problems for the most part, and people who favour the templates don't like to see advice about this kind of thing added to guidelines, so we've been somewhat hamstrung in trying to get the message out. SlimVirgin ^{TALK contribs} 02:31, 8 February 2010 (UTC)[reply]

Well, I was going to comment on this, but I just realized what's going to stymie any attempt to push towards this markup-tag defined reference - it will enforce a specific style on footnotes which, given everything else we know, is going to fly like a lead brick. Even proposing a user-pref selection of format, we'd infight over that like we did for date formats, for example. Getting this technical solution in place is going to require a lot of convincing of people that it is in our best goals to get over this "no common format" issue that currently persists; but if it is a cut down in page rendering time and thus overall CPU usage improvements, those benefits may outweigh the concerns.

I think the key thing out of all this is that we are going to have to standardize something to make editing easier for all editors, or otherwise be resigned to let the situation persist because we all insist on our own personal styles for referencing. Something has to give if we want to move from the status quo. --MASEM (t) 02:43, 8 February 2010 (UTC)[reply]

Rather than trying to move people to using this or that, what I'm hoping is to begin the process of gentle encouragement to editors simply not to place these long templates inside texts. This is what I hope might eventually emerge from this discussion, that FAC writers are made aware that these templates can be problematic, and perhaps a sea change will filter down from there. One of the difficulties is that so few people understand the technical issues, myself included. If we could clearly show that templates slow things down, and could get that message out, my guess is that many people would voluntarily adjust their habits, because none of us wants to create articles that others can't access. But convincing people of that when they don't understand the technical stuff is close to impossible. Still, a lot of useful technical information was generated here. Perhaps we can find a way to use it. SlimVirgin ^{TALK contribs} 03:14, 8 February 2010 (UTC)[reply]

The demo by rannṗáirtí anaiṫnid described in Wikipedia:Centralized discussion/Wikipedia Citation Style #Demo of specific proposal allows for different styles (the demo does both APA-ish and Vancouver system styles), but they are not implemented using templates (that's why it's so fast). Alas, this demo doesn't (yet?) work on Wikipedia because it requires changes to MediaWiki. Eubulides (talk) 06:39, 8 February 2010 (UTC)[reply]

Let me restate what I think the three main issues are here, as this will help lay a path for what needs to be done to arrive at some solution:

Full inline citations make it more difficult to parse through text than either brief spelled-out citations or reuse of named refs.
Template forms for citations are more confusing than spelled-out citations for newer editors particularly when used inline.
Template citations can slow down the rendering engine for WP.

Now, I will throw out these cautions:

Editors hate to standardize. Self-consistency within articles, yes, standardization across WP, no.
The templates do serve a useful function for editor to help identify what fields we want if they have not done significant referencing before (though admittedly they sometimes provide too many fields) and make it easy to format to a specific style of reference.

That said, I think an appropriate "plan of attack" would be as follows:

Get list-defined references as a preferred standard for articles. This does not necessarily mean forcing it as a requirement at FAC, but does mean getting the word out, seeing out bot editors to help with conversion and possibly upkeep, and the like. LDR is relatively new but addresses many of the past concerns about the weight of citation text in article bodies. This can be done in the immediate short term, requiring no Mediawiki changes, and only as a propaganda approach to sell the idea better.
Push for the suggested ref-tag expansion to embed the citation info. This will remove the template issue with both speed and readibility (to some extent). This would require a lot of expansion on the MediaWiki side to make sure that all the required fields are given to generalize the tag and to provide necessary style variations. There's more work here, but as the push for LDR progresses, this can come in on the back end to complete the process. --MASEM (t) 16:43, 9 February 2010 (UTC)[reply]

I suspect that (2) has a better shot at gaining acceptance, assuming it can be implemented. If the resulting appearance is the same, and if the ref-tag code has the same functionality and is about as easy to edit as the template code, surely editors will prefer that change, as it greatly improves performance. As for (1), not all editors prefer list-defined references. Some prefer keeping the citations in the text, as this makes it easier to cut and paste text (and associated refs) from one article to another, and to delete refs when the corresponding text is deleted. So that may be more of an uphill battle, though it certainly can't hurt to publicize this functionality. Eubulides (talk) 20:40, 9 February 2010 (UTC)[reply]

There is definitely no harm in working towards the expanded ref tag functionality. The issue with the first point is that we're balancing what experienced editors are accustomed to and what newer editors are going to be seeing, and to exactly which group are we catering for. This is close to, but not quite, the same issue of date linking and formatting, but there it was more based on the display of dates which affected all readers, not just editors. If, as SlimVirgin suggests, in-line cites is a significant barrier for newer editors to use and to some extent experienced editors, then it would suggest we want to make whatever barrier there is for getting editors involved in WP editor as easy as possible to surpass, and thus would suggest this change. I do recognize that there are issues with LDR for section editing or cross-page moves, and all that starting from the fact we don't have a WYSIWYG editor for WP. So if we all agree that we need to improve editing for newer editors, something is going to have to give on the experienced editor's side barring any great technical improvement in editing via MediaWiki software in the first place. --MASEM (t) 16:39, 10 February 2010 (UTC)[reply]