Template talk:High-use/Archive 1

Archive 1

About these templates

The {{intricate}} template has long been used for a similar purposes as these {{high-risk}} / {{high-use}} templates. But I have always felt that one doesn't really fit for high-risk / high-use templates, since it mostly talks about the code being intricate.

That a template is intricate is usually obvious when one looks at the code, no need to tell that. What is not obvious from just looking at a template and its code is if a template is used on many thousands of pages. So this template is a way of telling that.

Also, I have noticed that many admins are sloppy and don't use sandboxes, instead they try things out directly on protected widely used high-risk templates. Edit comments like "just testing" on templates used on 100,000 pages or more are all too common. This template points them to the /sandbox and /testcases, even if those subpages have not yet been created. This also means that non-admins are pointed to those subpages and thus can help out by experimenting in the /sandbox.

--David Göthberg (talk) 03:06, 14 May 2008 (UTC)

Wording

This discussion was moved here from David's talkpage:

Since you are a native English speaker and I am not, and you are not me and thus might see things I have missed: I would like some help to check and correct the text of some templates and their documentation that I have created lately. And if anything in their explanations is unclear ask me and I'll explain and thus I will know what might need a better explanation.

• {{high-use}} – I think this one should be pretty straightforward to understand, so mostly a language/wording check.

Looking forward for your comments on this!

--David Göthberg (talk) 02:36, 18 May 2008 (UTC)

Please take a look at User:Matthewedwards/Sandbox/Template:High-use. I re-worded the template slightly, and will take a stab at the documentation later tonight. -- Matthewedwards (talk · contribs · count · email) 23:24, 18 May 2008 (UTC)

I like your rewording of {{high-use}}. I think we should use your rewording pretty much as is, except two details: I think we should keep "server load" instead of "server bandwidth" since it is not about connection bandwidth but about CPU load. That is, every time you change a template all pages that use it has to be re-rendered. And perhaps (but just perhaps) we should say "please discuss any changes" instead of "please discuss any major changes" since for the really high-use templates I think even tiny changes should first be discussed. Put perhaps people will think it is too much to ask that? Or something like this is perhaps in between: "please discuss your changes". Or would that sound weird?

Thanks for taking a look and for the improved {{high-use}} text!

--David Göthberg (talk) 04:18, 19 May 2008 (UTC)

I wasn't sure if "Bandwidth" was correct, but I wonder if the majority of users will understand "server load" and what it means regarding every page being re-rendered? Personally, I don't like second-person usage, so instead of "your", how about "please discuss any changes"? -- Matthewedwards (talk · contribs · count · email) 06:34, 19 May 2008 (UTC)

I have now updated the text of {{high-use}} using a slightly modified version of your text.

And yeah, "load" can mean any kind of load. But since "bandwidth" would be untrue I prefer to use "load" even thought the word might seem less clear. I think we need to link to more explanation somewhere. There already are two pages explaining these things ( Wikipedia:Template test cases and Wikipedia:High-risk templates ) but their explanations are currently pretty bad and I didn't know how point to them in a nice way in this message box so I left that out for the moment. I am thinking we should perhaps instead add an explanation at the top of the message box's documentation about server load and why testing is a good thing before deploying (lots of good reasons for that). And then perhaps link to that explanation from the message box? This message box is only used on template pages so we are allowed to link to its own documentation like that.

--David Göthberg (talk) 18:44, 19 May 2008 (UTC)

Splitting ((high-use)) and ((high-risk))

This discussion was at first a part of the section above, but since it led to the splitting of the template into two templates I added this section title. --David Göthberg (talk) 22:09, 11 February 2009 (UTC)

I've removed discussion of "bandwidth" or "server load" from the wording. Firstly, while it may be a concern to a select group of editors who do lots of work on extremely high-profile templates, WP:PERF is not supposed to be something that we concern our users with, and not every change to a template considered to be in "high use" (which I've seen applied to templates with a hundred transclusions) is going to cause the server farm to fall over. Secondly, the major concern should not be performance anyway - it should be screwing up and having that negatively impact hundreds or thousands of articles. I've reworded the template to reflect this. Chris Cunningham (not at work) - talk 09:00, 19 November 2008 (UTC)

Chris Cunningham: You have misunderstood this message box in several ways. So let me explain:

1: We made this message box since we needed something to put on the very high-use templates. That is the top 20 or so templates. Those are used on about 300.000 - 3.000.000 pages each. We had frequent problems with admins doing fast repetitive experimental edits on them. Thus causing widespread disruption to page layouts. And even pushing up the job queue so much that we got noticeable effects such as editing got slower for everyone and category list updates got seriously delayed. (Back in 2007 such edits even caused server crashes. But the devs fixed that so now the servers at least do not crash anymore, instead things just get slow.) So this message box is not directed towards regular Wikipedia editors, instead it is directed to admins only.

2: "Server load" in this case does not mean bandwidth. In this case it mostly means server CPU load. If you do 10 edits within some minutes to a template that is used on half a million pages, then we do need to think about performance.

3: The sentence "The tested changes can then be added in one single edit to this template" is essential when you deal with the kind of fools like the admins that do 10 experimental edits in 5 minutes to a template used on 300.000 pages. They even have used comments like "Experimental edit, probably won't work" in their edit comments. So they knew they were experimenting, still they did it directly on the high-risk production templates. And this happened all the time. After we added this message box to the highest risk templates it seems such incidents have become much rarer.

4: Yes, the visible disruption to lots of pages is today probably the bigger concern, since the servers are more stable now. And that is what we meant with "disruption". And yeah, the text should perhaps be somewhat tweaked to reflect that.

5: I think that changes to templates that are visible on that many pages should be discussed before deployed. Thus we used the wording "please discuss any changes". You changed it to "consider discussing changes", which I think is a too soft statement.

6: You removed the line break after the first bold sentence. I find that less readable. The first sentence is the most essential part and works as a heading.

Thus I have reverted most of your edit.

--David Göthberg (talk) 03:31, 27 November 2008 (UTC)

That's fair enough; the new lower bounds on where the template should be used means that it doesn't apply to the templates I was really concerned about. I'll think about creating a new header for those. Chris Cunningham (not at work) - talk 08:42, 27 November 2008 (UTC)

Oh, that is a good idea! I agree, we do need a softer message we can use on many other templates to inform people that they can use the /sandbox etc. I am thinking something like this:

• For the really high-risk templates (top 100 or so), we can use the template name {{high-risk}}. (That is currently a redirect to this {{high-use}} template.) And {{high-risk}} can use the current pretty hard wording, and the current orange major warning "content" colour.
• For the rest of the high-use templates (down to say templates used on 2000 pages or so), we can use the template name {{high-use}}, and use the softer text you suggested in your edit to this template. I really like your softer wording, and for this level it would be perfect. And we can use the yellow minor warning "style" colour and the yellow warning triangle we often use for the yellow level.
• For regular templates we could make a template named something like {{sandbox reminder}} or so. I have to think a while for what is the best name for that. Name suggestions are very welcome. That one should probably use the default blue "notice" level and default blue (i) icon. That template should perhaps mostly be placed on templates were we expect or already get frequent edits, and on semi-complex templates. For the really complex templates we already have the {{intricate template}}.

I am not so concerned about exactly what bounds we choose between the three templates, I just mentioned some recommendations so people don't need to waste time thinking about it and can just follow the recommendations.

Chris Cunningham and anyone else reading this: What do you think?

--David Göthberg (talk) 10:19, 27 November 2008 (UTC)

Sounds good to me. Chris Cunningham (not at work) - talk 08:55, 5 February 2009 (UTC)

Y Done - I have done the split into {{high-use}} and {{high-risk}}. And I have updated all pages that transclude or link to these templates, that took about 100 edits, phew... And while doing that I noticed that people were already using {{high-use}} for many not so high-risk templates, so thanks Chris for coming up with the idea of making it two templates.

--David Göthberg (talk) 05:29, 12 February 2009 (UTC)

Excellent work. Thanks David. I've tweaked the wording on {{high-risk}} to be a little stronger now. Chris Cunningham (not at work) - talk 09:42, 12 February 2009 (UTC)

Thanks. And wowsers!, that is really strong wording, but yeah, probably appropriate for the {{high-risk}} message box.

How do you feel about the bounds I state in the docs for when to use these message boxes? (Currently 2000 - 100,000 for {{high-use}} and 100,000 and up for {{high-risk}}.)

By the way, one reason that I and others leave an empty line between the "<noinclude>" and the "{{documentation}}" in the template code is to make it easier (less error prone) when we cut and paste back and forth between a template and its /sandbox. (There are several reasons why we don't want to display the /doc page on /sandbox pages. And the sandbox often contains a {{template sandbox notice}} that we want to retain.) And that empty line only costs two bytes of template code, and that is in the noinclude area thus not affecting the size when transcluding the template. And some of us find it more readable since it keeps the actual template code and the "administrative" parts apart.

--David Göthberg (talk) 18:00, 12 February 2009 (UTC)

The boundaries sound good. We can always change them if we come up with better ones. As for the whitespace on the doc line, I normally leave a blank line just above the end of the template, as the mbox code ignores whitespace. But I'm sure you're aware that getting a programmer to use someone else's whitespace conventions is about as easy as getting dogs and cats to play together. :) Chris Cunningham (not at work) - talk 18:12, 12 February 2009 (UTC)

You mean leaving a blank line like this?

{{ambox
| text =
Lots of text and complex code.
                  <!-- The "blank" line. -->
}}<noinclude>
</noinclude>

Yeah I sometimes do that too, especially before the end braces of complex #switch cases. MediaWiki allows such a blank line before several uses of end braces. But we have a special tweak in the mbox's code to make them ignore whitespace around their "text" parameters.

And true about whitespace conventions. I knew it was a long shot, but I wanted to explain why I code like I do anyway. So far I have managed to convert many template programmers to use more whitespace! :))

--David Göthberg (talk) 18:42, 12 February 2009 (UTC)

Edit request

{{editprotected}} Can the link to Special:MostLinkedTemplates be replaced with Wikipedia:Database reports/Templates with the most transclusions, since this page offers more up-to-date information? PC78 (talk) 20:55, 14 May 2009 (UTC)

Page is not protected :) — Martin (MSGJ · talk) 21:46, 14 May 2009 (UTC)

Ah, well, there you go. PC78 (talk) 06:42, 15 May 2009 (UTC)

Update number of uses dynamically?

Is there some way the template parameter for number of uses can be updated dynamically from the data in Wikipedia:Database reports/Templates transcluded on the most pages? Set theorist (talk) 09:50, 26 March 2012 (UTC)

Should the High-risk message box be in includeonly tags?

There are several places where editors have surrounded the {{high-risk}} template by <includeonly>, tags, such as Template:!/doc (where I've since removed them), Template:WPBannerMeta/doc, Template:Class mask/doc, Template:Class/doc, Template:Class/icon/doc, Template:Importance mask/doc, Template:Importance/colour/doc, Template:Importance/doc Template:Reflist/doc, Template:Tl/doc, Template:Asbox/doc, Template:Navbox/doc. This seems unnecessary in light of this edit. Furthermore, I've changed the text of the message to make it clear that it is warning users about "any changes to this template" (rather than changes to the documentation). If anyone wants to clarify the wording further, or add a sentence about being bold with changing the documentation, that might be helpful, but it should be fine to remove the noinclude tags from uses of this template on documentation subpages. Set theorist (talk) 09:20, 18 April 2012 (UTC)

I removed the includeonly tag from many uses of {{high-risk}} but there are still more to do. Set theorist (talk) 09:05, 24 April 2012 (UTC)

This seems undesirable behavior. Partially, this template tracks the actual templates which have hundreds to millions of transclusions. We don't want to add /doc pages to those results, it would seem to me. --Izno (talk) 14:33, 24 April 2012 (UTC)

sandbox and testcases links

When this template is placed in a /doc subpage, it works fine whether that /doc subpage is viewed on the main template, or directly. However, there's a problem if the /doc subpage is viewed on the template's /sandbox, as with Template:WikiProject Aviation/sandbox: the links are to non-existent pages Template:WikiProject Aviation/sandbox/sandbox and Template:WikiProject Aviation/sandbox/testcases. Of the two, the second is the greater problem because having edited a /sandbox you will often wish to go directly to the /testcases subpage, and the red link may lead the editor into believing that no /testcases exists, and that they should create it; whereas it does in fact exist at Template:WikiProject Aviation/testcases.

I think that the code that generates these two links should detect that this notice is used the /sandbox and /testcases subpages, in the same way that it detects use on the /doc subpage. --Redrose64 (talk) 11:11, 20 August 2012 (UTC)

Special:TemplateSandbox

It seems like it would be helpful if these templates would mention the existence of Special:TemplateSandbox, which sounds really handy for testing modified templates on real pages. —SamB (talk) 18:37, 23 February 2014 (UTC)

Protected edit request on 4 April 2014

Change the link to a WM Labs link like this: Template:Tl, the database report is so stale (not updated more than six months ago). Eyesnore (pc) 20:45, 4 April 2014 (UTC)

Question: Which link would that be? --Redrose64 (talk) 20:58, 4 April 2014 (UTC)

Like this: http://tools.wmflabs.org/templatecount/index.php?lang=en&namespace=10&name={{urlencode:BASEPAGENAME}}, with plain link span. Wikipedia:Database reports/Templates transcluded on the most pages is too stale (not updated more than six months ago). Eyesnore (pc) 21:23, 4 April 2014 (UTC)

Not done for now: That link isn't in this template. Please sandbox your proposal at Template:High-use/sandbox. --Redrose64 (talk) 21:32, 4 April 2014 (UTC)

Recent merge

To editor Galobtter: just wanted to thank you for your majestic effort and work that resulted in the merge of template {{High-risk}} into this {{High-use}} template! Just awesome!  Paine Ellsworth  ^put'r there  12:37, 28 October 2018 (UTC)

Number of transclusions

I can see little benefit in displaying the number of transclusions a template has. Firstly whatever number is used is likely to be out-of-date very soon. It encourages pointless changes like this which have no benefit to the encyclopedia. And because it really makes no difference on whether a template has 1000 or 1000000 transclusions, you still need to be careful not to introduce errors. What do people think about removing this parameter? — Martin (MSGJ · talk) 09:03, 14 November 2016 (UTC)

Courtesy ping to User:Paine Ellsworth. I'm not criticising you personally, but rather the system that encourages you to make such edits. — Martin (MSGJ · talk) 09:14, 14 November 2016 (UTC)

Thank you, Martin. I use these and update them extensively in the rcats that need them, and I do this because I think {{High-use}} and especially {{High-risk}} have useful messages for template editors. I also think they might help deter vandalism of the templates, which is largely just test or ego edits. So I don't know if removing the parameter would change things much. The thing is, it seems like I'm always coming across templates with from 2K to 200K transclusions that do not even have one of these templates, so the parameter might help editors stay aware and abreast of the need to apply these, and when to apply them. They're not difficult to update nor do they take any time at all to do so.  Paine  ^u/c 10:22, 14 November 2016 (UTC)

I'm happy to accept that it is useful to warn editors that a template is high-risk. What I'm questioning is whether the number of transclusions is important, and whether it is worth your valuable time updating a number from 1216000 to 1221000. How will that additional 5000 transclusions make any difference at all? — Martin (MSGJ · talk) 08:58, 15 November 2016 (UTC)

To editor Martin: In a case like that the only difference (as long as the parameter exists) is accuracy, which for that many transclusions won't really become important again until the percentage begins to be used, such as at {{Navbar}}. And I'm not sure who else finds the actual figures useful behind the scenes, but it is clear that the higher the transclusions, the heavier the server load when a change is made, and the worse it is for the system when back-to-back edits are made.  Paine  ^u/c 21:27, 15 November 2016 (UTC)

What if he were to round the number to the first significant figure? That would provide enough information for editors, and would not require such frequent updating. — Martin (MSGJ · talk) 08:53, 16 November 2016 (UTC)

Examples now on /testcases — Martin (MSGJ · talk) 09:31, 16 November 2016 (UTC)

It should round to the second sig fig IMO. 1.5m rounding to 2m seems... not quite a valuable rounding. --Izno (talk) 12:34, 16 November 2016 (UTC)

That sounds good, for example at {{Navbar}}, that can be rounded from 7,150,000 up to 7,200,000. Since the actual count at present is 7,159,315, the plus sign can't be used, as in 7,150,000+. It should read 7,200,000 without the plus sign, and these new details (if this is the way we decide to take it) should of course be clearly described in this template's documentation, as well as at {{High-risk/doc}}.  Paine  ^u/c 15:03, 16 November 2016 (UTC)

I have updated the sandbox to use 2 significant figures — Martin (MSGJ · talk) 22:23, 16 November 2016 (UTC)

I'm still seeing editors somewhat compulsively updating the documentation each time the template count increases by a few percent. I should be able to put together a bot to create a lua-formatted file similar to Wikipedia:Database reports/Templates transcluded on the most pages, and then use a module to automatically fill in the transclusion count. Would there be any interest in this? --Ahecht (TALK
PAGE) 18:56, 5 June 2019 (UTC)

• I think it would be a useful bot. Peter coxhead (talk) 06:57, 6 June 2019 (UTC)

Suggestion

Where this template says "all pages" it would do well to specify "all English Wikipedia pages" 2A01:CB08:62:8100:2453:4427:E262:8786 (talk) 18:36, 21 July 2019 (UTC)

The edit would need to be made here: Template:High-use/num. 2A01:CB08:62:8100:2453:4427:E262:8786 (talk) 18:44, 21 July 2019 (UTC)

Why? --Izno (talk) 03:22, 27 July 2019 (UTC)

Because editors can have the mistaken impression that their edits will affect other language wikis when in fact that will not be the case. 2A01:CB08:62:8100:18A6:D391:A202:E0F1 (talk) 13:23, 27 July 2019 (UTC)

I doubt highly that editors of high-use templates will think that. --Izno (talk) 15:15, 27 July 2019 (UTC)

Switch to automated transclusion count

I have been running Ahechtbot Task 6 for a while now, creating a database of the transclusion counts of templates with over 2000 transclusions. I have also created a version of this template at {{High-use/sandbox}} that uses Module:Transclusion count to automatically retrieve the transclusion count values from the database and uses them instead of the manually input value wherever possible. I have implemented the sandbox template on Template:R from sort name/doc, Template:WikiProject Albums/doc, Template:Taxobox/doc, Template:Small Solar System bodies/doc, Template:Portal/doc, and Template:Memory Alpha/doc, which should represent a variety of use cases (5-digit count, 6-digit count using "+" notation", 6-digit count using "approximately" text, 4-digit count, 7-digit count with percent, and a template that is not indexed by the bot). The template falls back to the manually input value if it cannot find the template in the database.

Any objections to switching over the main template to use the module to automatically update counts? --Ahecht (TALK
PAGE) 20:08, 5 October 2019 (UTC)

How expensive is this? That is, how frequently will it force pages to be reparsed? --Redrose64 🌹 (talk) 07:48, 6 October 2019 (UTC)

@Redrose64: The database is processed weekly, but it is divided between 27 pages (one for each letter of the alphabet and one "other") and values are rounded off to two or three significant figures (depending on the total number of digits). Only templates that share a first letter with a template whose usage has changed enough to change those significant figures could trigger a re-parsing. Worst case, if every use of the template needed to be updated, we're only talking about 2700 re-parsings a week (and these pages would've been re-parsed at most every 30 days if we just waited for the page to be edited or the parser cache to expire). --Ahecht (TALK
PAGE) 17:02, 6 October 2019 (UTC)

It's been a couple of months of the bot running smoothly and no objections, so I'm going to go ahead and implement the module in the live template. --Ahecht (TALK
PAGE) 21:39, 2 December 2019 (UTC)

@Ahecht: could the bot add to the /doc pages of the templates it takes care of either a hidden comment or a |bot=yes so editors will know not to update figures? (and when doing that, also remove the manually entered value in the template). I was going around in circles trying to understand why the values aren't changing no matter what I do. --Gonnym (talk) 15:23, 11 January 2020 (UTC)

@Gonnym: The bot doesn't have permission to update any pages other than those under Module:Transclusion_count/data/, so it cannot modify template transclusions or doc pages. In general, editors shouldn't ever be updating the values manually, as it wastes time and resources and clogs up watchlists. At some point, we might want to have it ignore manually entered values altogether and just use "many" in those cases. --Ahecht (TALK
PAGE) 16:14, 12 January 2020 (UTC)

I wouldn't object to removing the option to manually update numbers and just handle it with "many" or bot figures, but I do object to this current situation, where an editor has no idea why something isn't working. I'm sure getting a bot approval to do a one-time update the "bot-owned" template /docs isn't something that would be hard to get approved. Also, it seems the bot stopped working altogether over 2 weeks ago. Leaving Template:Link language with very inaccurate numbers. --Gonnym (talk) 09:39, 15 January 2020 (UTC)

I would also support going over to bot only transclusion counts. ‑‑Trialpears (talk) 16:05, 16 January 2020 (UTC)

If the database replica servers are too busy the bot won't run, but it will get more aggressive in trying to run as the values get more and more out of date. I honestly don't see any issue with values being a few weeks old. If a template suddently goes from 0 to 100,000 transclusiosn or something, you can always set parameter 1 to "risk" to make the warning look more serious instead of waiting for it to update. I have a sandbox a version of the template that will say when it is using the bot values, see Template:High-use/testcases. --Ahecht (TALK
PAGE) 16:31, 16 January 2020 (UTC)

I'd drop the small text, I'm really not old and that size really hurts my eyes. --Gonnym (talk) 14:03, 20 January 2020 (UTC)

We need a smaller boat

I suggest we add an option to reduce this template to a one line only. First, that should do for the admin/TE who embarks on editing (TE protection should be there of course). Second, those six lines of verbosity it takes today is distracting from the main function of a documentation page, and is not even addressing the visiting editor. (trigger example: it was re-added [1]). -DePiep (talk) — Preceding undated comment added 18:27, 22 September 2014 (UTC)

Formatting tweak

I'd like to wrap the bottom line, Transclusion count updated by Ahechtbot., in a small tag, since that information is not as important as the count itself or the instructions. Would that be alright? {{u|Sdkb}} ^talk 03:06, 7 August 2020 (UTC)

Hmm, I'm not sure how to code Scribunto here. Ahecht? {{u|Sdkb}} ^talk 22:59, 28 August 2020 (UTC)

@Sdkb: You would make it small by changing line 57 of Module:High-use to

	local bot_text = "\n\n----\n<small>Transclusion count updated by [[User:Ahechtbot|Ahechtbot]].</small>"

but see the comment from Gonnym above before making any changes. --Ahecht (TALK
PAGE) 22:44, 29 August 2020 (UTC)

Hmm, how small was it before? The general practice in a whole bunch of templates seems to be that it's okay to use the small tag, so long as it's not on top of something else. MOS:FONTSIZE I think says not to go below 85% of the default. But I'd like there to be some reduction/de-emphasis, since for usability reasons it's important to have a visual hierarchy highlighting what's actually important. {{u|Sdkb}} ^talk 22:50, 29 August 2020 (UTC)

It was too small to read. I don't see a real issue here with having it normal size. What difference is this one sentence in the page length compared to the ease of use? --Gonnym (talk) 23:01, 29 August 2020 (UTC)

Gonnym, it's a usability issue. 99% of people coming across this template aren't going to care which bot does the count updating. What they care about is that the template is high-use, and what that means for how they edit (i.e. the stuff above the line). Banner blindness is a huge problem on Wikipedia that myself and others at WikiProject Usability are trying to address, and having non-important information at normal size on templates contributes to it. Every template should be long enough to communicate what's important and no longer. There's really no need to mention Ahectbot at all, so long as it's in the documentation here in case any issues arise. But for the sake of giving credit, I suggested it be minimized rather than removed. {{u|Sdkb}} ^talk 01:16, 30 August 2020 (UTC)

Not sure what usability issue I was supposed to see in the links you posted. Banner blindness says people ignore text in banners, so if they ignore the text, what does it matter? Anyways, the previous text was too small, if you can make it not as small and still readable give it a shot. I'll be completely opposed to remove it, and it's not for credit reasons, but because it's the only place that gives indication to why the numbers aren't being updated when you write them. I saw first hand people editing over and over trying to "purge" or get it to somehow work. And if you think people ignore banners, people ignore template /doc even more. --Gonnym (talk) 08:24, 30 August 2020 (UTC)

@Sdkb: Here is how it looked with the <small>...</small> tags:

--Ahecht (TALK
PAGE) 16:31, 4 September 2020 (UTC)

I would remove the text entirely. If someone needs to know how it's updated they can consult the documentation. --Izno (talk) 02:14, 30 August 2020 (UTC)

@Izno: I can remove the "Ahechtbot" part of the line (as Gonnym said above), but the template needs some way to indicate that it is using an automatically generated value and will ignore a manually input one (that line is generated dynamically if and only if an automatic value is found in the database). For example, I could do "Tranclusion count updated automatically (see documentation)." --Ahecht (TALK
PAGE) 16:24, 4 September 2020 (UTC)

... Yes, that still falls under my original comment. Users can do lots of things with templates and when they find those things don't work, they consult the documentation. This template isn't different in that regard. If you really want them to know not to put it in, the template should raise an error when automated results are found and there is a number in the template output, rather than subtly display a message about the automation. --Izno (talk) 17:50, 4 September 2020 (UTC)

Typo

I believe that "Tranclusion" in the template text should be "Transclusion" instead. In accordance with the warning displayed on the template page, I'm mentioning this typo here, instead of just going ahead with the change. Let me know if it can be fixed. Thank you. Toccata quarta (talk) 10:54, 12 September 2020 (UTC)

Toccata quarta,  Fixed --Trialpears (talk) 11:41, 12 September 2020 (UTC)

Maths error

The |all-pages=yes code seems to be broken, and is implausibly low results. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:08, 15 September 2020 (UTC)

Are you saying that the value of 5% shown here is incorrect? If so, what would be a more plausible value? If not, where are you seeing an incorrect figure? --Redrose64 🌹 (talk) 22:19, 15 September 2020 (UTC)

My mistake; I was confusing "pages" with "articles". Apologies. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:00, 16 September 2020 (UTC)

formatNum is being applied to a non-numeric string

When |1=risk and the bot doesn't have a count, the phrase "a very large number of" is inserted. It appears that later in the code, formatNum is applied to that text, which is usually a number. That process puts pages like {{BLP editintro}} into the new tracking category Category:Pages with non-numeric formatnum arguments, since formatNum is supposed to be applied only to pure numeric values.

I don't know Lua well enough to modify this module's code, but can someone please try to find a way to avoid applying formatNum to this prose? I see that "approximately" is added a a post-processing step; perhaps "a very large number of" could be inserted at that point as well, instead of being inserted before number formatting is attempted. – Jonesey95 (talk) 14:03, 29 September 2020 (UTC)

The issue is in Module:Transclusion count at

	-- If database value doesn't exist, use value passed to template
	if return_value == nil and frame.args[1] ~= nil then
		local arg1=mw.ustring.match(frame.args[1], '[%d,]+')
		return_value = tonumber(frame:callParserFunction('formatnum', arg1, 'R'))
		frame.args["nobot"] = true
	end

Seeing if I can sort. --Izno (talk) 15:01, 29 September 2020 (UTC)

Nope, have not managed to sort it in either of the two places it might be. Ahecht? --Izno (talk) 15:24, 29 September 2020 (UTC)

I don't think that's it, but I could be wrong. formatNum with "R" does not appear to cause this category to be applied. – Jonesey95 (talk) 16:23, 29 September 2020 (UTC)

True. {{formatnum:1000 million}} and {{formatnum:1,000 million}} both put a page into Category:Pages with non-numeric formatnum arguments but {{formatnum:1000 million|R}} and {{formatnum:1,000 million|R}} do not. The comma makes no difference, it's purely the absence or presence of the |R. --Redrose64 🌹 (talk) 20:16, 29 September 2020 (UTC)

Does that make it a bug? Not sure what R does... --Izno (talk) 21:01, 29 September 2020 (UTC)

R is meant to strip commas rather than add them. Behaviour described seems expected. Strings containing commas should probably not be sent to formatnum, although a non-comma string to formatnum with R doesn't seem to hurt. ProcrastinatingReader (talk) 21:41, 29 September 2020 (UTC)

Not just meant to, but actually does so:

• {{formatnum:1000000}} → 1,000,000
• {{formatnum:1,000,000}} → 1,000,000
• {{formatnum:1000000|R}} → 1000000
• {{formatnum:1,000,000|R}} → 1000000

This is documented at H:MW#Formatting. --Redrose64 🌹 (talk) 22:19, 29 September 2020 (UTC)

Have updated the module with something that shoudl fix it. arg1 regex only returns digits and commas, non-digit strings will be empty - empty check on string should fix issue. ProcrastinatingReader (talk) 21:26, 29 September 2020 (UTC)

I'm not sure which module is being referenced here. Module:High-use has not been modified, and {{BLP editintro/doc}} is still in the relevant error-tracking category. – Jonesey95 (talk) 22:52, 29 September 2020 (UTC)

Relevant lines are 27-33. --Gonnym (talk) 23:17, 29 September 2020 (UTC)

ProcrastinatingReader, if you are referring to this edit of Module:Transclusion count, see the discussion above regarding the "R" option. I don't think Module:Transclusion count is causing formatnum to receive badly formatted text.

Gonnym, it sounds like you are referring to lines 27–33 of Module:High-use. I agree, and would also include lines 10–15. Can the functionality in lines 10–15 be invoked after the number is formatted? Maybe the logic could be reversed as follows: "if 'count' is not nil, do what is currently in the 'else' statement (lines 16–39); otherwise, run the code that is currently in lines 10–15." [edited to add: On second thought, I don't see how lines 27–33 are involved, based on my understanding of if/else/end statements. I'm confused.] – Jonesey95 (talk) 00:21, 30 September 2020 (UTC)

Jonesey95, my bad. Added a truthy check now (reason required: match returns nil not empty). try now - cat gone from doc. re your comment above, I believe main offender in that module is line 6/62, but doesn't really matter for fix I think - different module. ProcrastinatingReader (talk) 00:42, 30 September 2020 (UTC)

ProcrastinatingReader, thanks for digging into and fixing something that I clearly do not comprehend fully (hence my "I could be wrong" weasel disclaimer above). And Izno, it looks like you were right after all. Thanks to your both for your attention to this trivial detail. – Jonesey95 (talk) 02:08, 30 September 2020 (UTC)

Is there a bot applying this template as needed?

I noticed that it does not appear at {{Infobox song}}, despite that template appearing on close to 70,000 pages. {{u|Sdkb}} ^talk 02:47, 23 December 2020 (UTC)

Suggestion: merge with {{Used in system}}

This template is nearly word-for-word identical to {{Used in system}}, so I have attempted to merge them:

Does this seem like a good idea? The aim is to reduce banner blindness by not putting two near-identical banners on top of each other as normally happens with this template and {{Used in system}}. Can open a TfD later if required. User:GKFX^talk 12:52, 3 April 2021 (UTC)

Not a fan since this would make {{used in system}} a lot easier to miss. I think a lot of template editors including me more or less disregard {{High use}} since it's basically always a concern and the same process apply regardless of there being 1000 or 100 000 transclusions. {{Used in system}} on the other hand may very well require extra step a normal high use template does not. If anything I would just remove or significantly shorten {{High use}}. --Trialpears (talk) 13:00, 3 April 2021 (UTC)

That makes sense. I've gone over it to make it look more like {{Used in system}}, with the distinct icon and familiar wording in bold, and de-emphasised the transclusion count. User:GKFX^talk 17:47, 3 April 2021 (UTC)

I suppose that's an improvement, but having two visually distinct banners makes it very apparent that this template has something different going on. Using your merged version as {{Used in system}} which would retain the distinct nature of used in system but remove the need for two mostly redundant banners. --Trialpears (talk) 18:01, 3 April 2021 (UTC)

Wrong calculation somewhere?

The current version of {{Cite book/doc}} currently states:

• This template is used on 1,350,000+ pages, or roughly 3% of all pages (as seen in: {{High-use|demo=Cite book}})

Seems like that figure should be more like "roughly 20%" of all pages, no? Thanks, Mathglot (talk) 03:26, 22 June 2021 (UTC)

@Mathglot, since there are approx. 6 million articles, I assume "pages" in this case means total pages, not just article pages. That's not really the right comparison group, as I imagine cite book is used quite rarely outside mainspace. {{u|Sdkb}} ^talk 03:37, 22 June 2021 (UTC)

@Sdkb:, oh, I never even considered that possibility. Wow, are there really another 39,000,000 pages that aren't article pages? If there isn't already a {{High-use|space=articles}} param, maybe I'll add one. Mathglot (talk) 03:44, 22 June 2021 (UTC)

That would work. It might be slightly better to get an article-space count measured by the bot and then automatically display the right warning (e.g. >90% in article space gets the article-space warning.) User:GKFX^talk 12:08, 22 June 2021 (UTC)

This is irritated me for quite a while so I have done something about it:

{{high-use |demo=Module:Citation/CS1 |no-percent=yes}}

—Trappist the monk (talk) 23:33, 3 August 2021 (UTC)

Assuming it works, that looks good to me! {{u|Sdkb}} ^talk 03:12, 4 August 2021 (UTC)

Do you have any evidence that it doesn't work?

—Trappist the monk (talk) 10:24, 4 August 2021 (UTC)

Well about half a million of those transclusions aren't actually in articles. Ahecht could probably quite easily get mainspace figures as well though. --Trialpears (talk) 10:27, 4 August 2021 (UTC)

I think I could change the query to limit it to transclusions in article space by adding WHERE tl_from_namespace=0 to the SQL query, but I'm not sure how that handles recursive transclusions (where a template is used in a template that is used in an article). I'm doing a one-time run of the bot into subpages of Module:Transclusion count/sandbox/data/ with that option set, it should populate those pages in the next few hours. I did a test run in quarry on a much smaller wiki, and adding that option tripled the amount of time the query took, so it may or may not time out. If it does work, I'd then need to figure out how to to integrate both the article and total numbers into the regular data sets, and how to do it in such a way that I don't temporarily break the existing template. --Ahecht (TALK
PAGE) 02:19, 5 August 2021 (UTC)

Update: The database query for all namespaces normally takes less than 10 minutes to run. The one that tries to narrow it down to articles ran for three hours before timing out. --Ahecht (TALK
PAGE) 05:12, 5 August 2021 (UTC)

Trappist, I was just saying that I hadn't checked any of the technical aspects of the change, but on an editorial level, I support it. {{u|Sdkb}} ^talk 18:28, 4 August 2021 (UTC)

Nicely done, Trappist; thanks! Mathglot (talk) 10:51, 4 August 2021 (UTC)

So my edit has been reverted with the edit summary this would present very misleading data, as it would be comparing the number of transclusions on all PAGES against the number of ARTICLES. Wait to implement until there is actually a data source to match. Given that the experiment to create a data source to match times out, and, frankly, because the number of template or module transclusions in all pages is really a meaningless number, I have implemented |no-percent=yes to suppress the ", or roughly n% of all pages." portion of the rendering. My example above is modified to show this.

—Trappist the monk (talk) 12:06, 5 August 2021 (UTC)

@Trappist the monk: It might be better to just undo Special:Diff/1016519658. Prior to GKFX's edit four months ago, it only displayed the percentage if you explicitly set |all-pages=yes. I don't think there ever actually was a consensus to turn the percentage on as the default. --Ahecht (TALK
PAGE) 14:11, 5 August 2021 (UTC)

That would be ok but instead of:

if frame.args["all-pages"] and frame.args["all-pages"] ~= "" then

I would write:

if yesno (frame.args['all-pages']) then

|all-pages=no should not show the all pages text.

—Trappist the monk (talk) 14:35, 5 August 2021 (UTC)

Fair enough. It was originally written that way to emulate the behavior of the old non-lua template, which took any value to mean that it should use a percentage. I'll set the default to true, so that something like |all-pages=✔ would still work if someone used that with the old template, but |all-pages=no would successfully turn it off. --Ahecht (TALK
PAGE) 14:49, 5 August 2021 (UTC)

Yeah, ok, but if you are going to do that then the other no synonyms false, 0, etc should not be recognized as yes. Do that, and I'm happy.

—Trappist the monk (talk) 15:15, 5 August 2021 (UTC)

@Ahecht: Personally I think this template should aim to be as automatic as possible rather than being something that needs to be manually set to say the right thing. If a template is used on some percentage n of all articles then that is worth noting to give a sense of scale of the page count. If it's less than n, it’s not worth saying. That’s a decision the code can make. That said, I picked n = 1 arbitrarily; I’d rather see n changed to 5 or something than have a manual parameter. Or switched off altogether if it's too meaningless. User:GKFX^talk 17:03, 5 August 2021 (UTC)

High-use files

I think files should have a similar high-use template, as we do with templates (tagging files used in more than 5000 pages), and files that are used in more than 150k pages should be locally uploaded and protected. In the file description page, it would say: This file is used by more than 5,000,000 pages, ... Thingofme (talk) 14:53, 26 January 2022 (UTC)

all-pages/no-percent parameter

I noticed that this edit removed the all-pages parameter and this one re-added the equivalent (negated) no-percent. Is there an actual use case for the parameter or should it be removed after all?

The Wikitext side of the template and the test cases haven't been updated for either revision and still pass on all-pages.

Searching all template Wikitext, there are currently 35 uses of all-pages (all with the value 'yes') and 0 uses of no-percent.

--wqnvlz (talk | contribs) 10:55, 22 March 2022 (UTC)

TemplateData - no idea what the all-pages and system parameters are supposed to do, might need someone who knows what they do to fix that

Ok so I've attempted to add some TemplateData to this template's documentation, but I encountered a slight problem with the all-pages and system parameters; that problem being that I have no idea what they're supposed to do, and the existing documentation doesn't mention them at all. If someone here does know what purpose they serve, you might want to quickly edit the TemplateData to add some info about what they actually do. 🔥HOTm̵̟͆e̷̜̓s̵̼̊s̸̜̃🔥 (talk) 15:10, 23 August 2022 (UTC)

@HotMess: The "all-pages" parameter was removed a while ago from the underlying module so I have removed it from the template and will remove it from documentation too. The "system" parameter on the module is used to implement Template:Used in system, but I can't remember why I made it available from this template and it doesn't seem like a good idea. User:GKFX^talk 22:25, 23 August 2022 (UTC)

I'll be honest with you, I'm not entirely sure what most of those words in that last sentence actually mean 😅. Thanks for taking care of the all-pages documentation, but should system be marked as deprecated within/removed from the TemplateData, or just left undocumented as-is? 🔥HOTm̵̟͆e̷̜̓s̵̼̊s̸̜̃🔥 (talk) 01:32, 25 August 2022 (UTC)