Wikipedia:Village pump (technical)

(Redirected from Wikipedia:VPT)
 Policy Technical Proposals Idea lab WMF Miscellaneous 
The technical section of the village pump is used to discuss technical issues about Wikipedia. Bug reports and feature requests should be made in Phabricator (see how to report a bug). Bugs with security implications should be reported differently (see how to report security bugs).

If you want to report a JavaScript error, please follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk. Discussions are automatically archived after remaining inactive for five days.

Help needed with saving pages to a web archive Edit

I would like to save several pages of handball leagues to a web archive, but only partly succeed to do so.

The "fixtures", "results", "standings" and "matrix" (mix of fixtures and results) of those leagues can be viewed on the website of the Netherlands Handball Association (NHV).

As exampla I'll take the "Heren BENE-League":

  1. Goto
  2. Select "PROGRAMMA/UITSLAGEN/STANDEN" (Fixtures/Results/Standings) from the menu in the top black bar.
  3. Select "POULES" from the orange oval.
  4. Select the "Zaal" (indoor) entry from the "NHV Landelijk" line/section.
  5. Select the "Heren BENE-League" entry.
  6. from the "Poule BENE 23-24" line/section select either
    1. "Programma" (Fixtures) or
    2. "Uitslagen" (Results) or
    3. "Standen" (Standings) or
    4. "Matrix"

After the last selection you get the related info. You can switch between "Programma" (Fixtures), "Uitslagen" (Results), "Standen" (Standings) and "Matrix" info by selecting the corresponding tab. Don't use your browser functionality to go to the previous page in your browsing history (left arrow button before the address bar), it will result in having to go thru the menu all over again.

To my knowledge none of these pages can be saved on a web archive service like Wayback Machine or Archive.Today. This is caused because these pages are retrieved using HTTP POST requests contrary to HTTP GET requests.

These pages can also be retrieved using CURL on the command line:

Page CLI command
"Programma" (Fixtures)" curl -s -X POST -d "page=poules&tab=programma&params[]=XXX&params[]=35670" ""
"Uitslagen" (Results)" curl -s -X POST -d "page=poules&tab=uitslagen&params[]=XXX&params[]=35670" ""
"Standen" (Standings) curl -s -X POST -d "page=poules&tab=standen&params[]=XXX&params[]=35670" ""
"Matrix" curl -s -X POST -d "page=poules&tab=matrix&params[]=XXX&params[]=35670" ""


XXX A display string, which is not really relevant. For the league above it would be "NHV+Landelijk+^%^7C+Zaal+^%^7C+Heren+BENE-League+^%^7C+BENE+23-24" ("NHV Landelijk | Zaal | Heren BENE-League | BENE 23-24") without the double quotes.
35670 the ID of the league. Other leagues are e.g. 35674 for the "Heren Eredivisie" (Men Eredivisie) league or 35715 for the "Dames Eredivisie" (Women Eredivisie) league.

There's a undocumented way the request these pages in your browswer where an HTTP GET request is used:

Page URL
"Programma" (Fixtures)"
"Uitslagen" (Results)"
"Standen" (Standings)

Using these URL's clearly shows the effect of the display string XXX.

Changing 35670 into 35674 or 35715 will give the info for the other leagues mentioned above.

Using these URL's, the pages partly can be saved in a web archive.


Web Archive Comment
WayBack Machine None of the pages can be saved. WayBack cannot handle the "#" (anchor indicator) in the URL. Basically everything after the "#" is ignored.
Archive.Today Creates a correct copy for the "Programma" (Fixtures) and "Uitslagen" (Results) pages. However not for the "Standen" (Standings) and "Matrix" pages.

It's exactly the 2 pages I'm mostly interested in, for which it's not possible to create a web archive copy. End of season the "Programma" (Fixtures) is empty anyway. The "Matrix" page only contains results at the end, and I prefer the matrix representation of the results above the list presentation on the "Uitslagen" (Results) page.

At the start of a new season the pages of the previous season are no longer available. Hence I would like to have a web archive copy to use as reference.

Does anyone have an idea how to create a web archive copy of the "Standen" (Standings) and "Matrix" page. I don't care which method and web archive is used, as long as I get a proper web archive copy.

Any help is appreciated. TIA.

--Sb008 (talk) 17:04, 6 September 2023 (UTC)Reply[reply]

You will need to encode the URL. '#' is '%23'. I'v done one, leave you with the remainder (Wayback). Neils51 (talk) 13:38, 9 September 2023 (UTC)Reply[reply]
@Neils51: Thanks for your effort tp come up with a solution. Unfortunately it doesn't give the desired result. I tried it for a different league, 35715=Women Eredivisie.
Page URL Archive URL Result OK Shows all (layout), except for the actual data.
So, with HTML URL Encoding I get the same results on WayBack Machine as on Archive.Today without encoding. Actually, I normally already use encoding because instead of "XXX" I mostly use "%20".
Furthermore, there're 2 additional "problems" with using WayBack Machine suitable UEL's:
  • Browsers can't handle the encoded URL's, "%20" instead of "XXX" is no problem, but "%23" instead of "#" results in an HTTP 404 error.
  • Wayback Machine has issues anyway, too often it seems to make a copy, but you get the message "There was a delay in registering this snapshot with the Wayback Machine. You may be redirected to a previous version right now. This snapshot will be available later." And it seems to me that the snapshot never gets registered and you're always redirected.
Once again, thanks for your effort.
--Sb008 (talk) 11:26, 12 September 2023 (UTC)Reply[reply]
Keep open. Neils51 (talk) 22:32, 16 September 2023 (UTC)Reply[reply]

User:Sb008: Does this work? has more info. They use an advanced archiving engine that can save difficult UIs. We are supposed to use the long format URL, but when I try it, it doesn't display correctly. In which case you'll probably need to use the short form and add a {{cbignore}} to keep the bots from converting it to long form, and add a wiki comment that short is required.

If that is still not sufficient, Help:Archiving_a_source#Conifer will probably work, but, you have to sign up for a (free) account, and have limited though generous disk space, and need to use {{cbignore}} for the links otherwise IABot will delete them. -- GreenC 05:22, 19 September 2023 (UTC)Reply[reply]

@GreenC and Neils51: With I get the exact same results as with Archive.Today and with WayBack Machine + URL encoding, The "programma" (fixtures) and "uitslagen" (results) pages archive fine, the "standen" (standings) and "matrix" pages not.
When I archive the pages with Conifer, the archived copies are placed into a "collection" linked to your account. Opening the archived copies from the container gives a correct result for all 4 pages. However when I open any of the 4 archived copies from a browser window (outside Conifer), e.g. Matrix I see 2 arrows forming a circle (similar indicator as an hourglass in other applications), turning forever and I never get the info it's about.
When selecting the pages at using the 6 steps as mentioned at the start of my initial post and I start pushing "ctrl"+"+" (grow screen), it's not a problem on the "programma" (fixtures) and "uitslagen" (results) pages. However on the "standen" (standings) and "matrix" pages, the actual data will disappear. It depends on your screen size how often you'll have to push "ctrl"+"+". You get the same visual effect as for the archives of those pages. Pushing "ctrl"+"-" (shrink screen) will result in the actual data reappearing. Unfortunately you don't get the same effect on the archives.
--Sb008 (talk) 21:57, 19 September 2023 (UTC)Reply[reply]
When archiving at Conifer, try clicking around the site while the page is saving as it will save content interactively. Click on the tabs, scroll up and down the data windows, etc.. could feed it more data. Conifer can't save everything, in which case, that's really the best archive service for hard websites, some site may not be archiveable. You could contact the website owner who may be interested to learn this. -- GreenC 02:46, 20 September 2023 (UTC)Reply[reply]
@GreenC and Neils51: Although the pages are generated based on the league ID, the final result is pretty static. Therefor I cannot click around, cause it would result in getting another page, one I'm not interested in. Only thing I can do is scroll up and down and that doesn't change the outcome.
Meanwhile I discovered something else. I mentioned that opening archived copies (links) from the container gives a correct result contrary to opening them from a browser window. I also noticed that when opening the pages from the container, I correctly get the 4 tabs "Programma", "Uitslagen", "Standen" and "Matrix". I can even switch between those tabs and for each of them I get the correct actual data. When I open any of those pages from a browser window, I don't get those 4 tabs but instead only 1 tab ("clubs"). "Clubs" can be a valid tab, however not at this stage. If I go back to the steps mentioned at the beginning of my initial post, "clubs" is the default tab after step 2.
I did some more experimenting, In Conifer, if I enter the URL, which brings me directly to step 2 of the 6 steps, and I continue with step 3. At step 6 I choose any on the 4 options. The saved archive I get then, allows me to retrieve all info from the saved archive. If after step 6, I select "CATEGORIE", I go back to step 5. Next instead of selecting "Heren BENE-League", I can select "Heren Senioren (L)" and next I can select again any of the 4 choices "Programma", "Uitslagen", "Standen" or "Matrix" for e.g. "Poule ED HS-A" (Eredivisie Men) and all info of this league is saved into this archive as well. Again choosing "CATEGORIE" and next "Dames Senioren (L)", and again 1 of the 4 choices for e.g. "Poule ED DS-A" (Eredivisie Women) results in that league to be added to the archive as well. This way, I end up with an archive with info of 3 leagues in it (3 league archive). I assume I could save any number of leagues in this way into an archive. When navigating to a league which wasn't selected during the archive process at Conifer, it results in the in the 2 arrows forming a circle, circling forever (hourglass). Which is kinda logical.
The disadvantage of this method is that I would have to add instructions to a reference with an archive URL like this (end of season before the info is removed on the NHV site.) But it beats all other options so far. The only other option I can think of is making screenprints of the league pages, put those images somewhere where I can create a free website and create archives from those pages, which should be no problem because pages only contain images.
Although this isn't the prefect solution, I'll close those thread for now. I would like to thank both of you for the your input and if you can up with a better option, feel free to contact me directly.
--Sb008 (talk) 20:20, 22 September 2023 (UTC)Reply[reply]

Dark Mode and hardcoded colors Edit

Is there any way, for example, in lists of colors (like in iPhone 15) to exempt certain colors/markup from being converted to "dark mode"? It's undesirable to see a dark hue swapped for a light hue (or vice versa) on something that should be showing color accuracy. Like a CSS class or something that exempts an element from being modified? —Locke Coletc 03:41, 13 September 2023 (UTC)Reply[reply]

The class .mw-no-invert prevents color inversion. – SD0001 (talk) 04:53, 13 September 2023 (UTC)Reply[reply]
@SD0001 Thank you! That got it. Implemented at iPhone 15 Pro and iPhone 15 on just the table color cells and it survives swapping back and forth. =) —Locke Coletc 04:05, 14 September 2023 (UTC)Reply[reply]
Actually, follow-up: it indeed works on the desktop site when using the dark mode gadget. However, in the Wikipedia app (on iOS anyways) dark mode seems to make the colors show as black. In light mode, the colors show as expected. Is there a way to suppress the Wikipedia app from changing the colors? —Locke Coletc 05:31, 18 September 2023 (UTC)Reply[reply]
Oddly, it also doesn't seem to work when attempting to apply it to the colored cells in Interpol notice (in the notice types section). Even when applied to the table itself, every other cell remains light in dark mode, but the colors are still reversed. Very bizarre. —Locke Coletc 04:36, 20 September 2023 (UTC)Reply[reply]
Consider filing a Phabricator ticket and tagging it "Wikipedia-Android-App-Backlog". Consider mentioning .mw-no-invert and suggest that they program the app to use it. –Novem Linguae (talk) 23:00, 25 September 2023 (UTC)Reply[reply]

Vector 2022. Sticky page title covers top 2 lines of sticky table headers Edit

Vector 2010 has no sticky title to cover up sticky table headers. Is there any way to fix this in Vector 2022? Or can it be fixed locally in the template:

See: Help:Table#Tables with sticky headers. That section has been much updated.

Can the sticky table header in the template be made to have priority over the page title bar? People can always go back to the top of the page to get that page title bar.

In Vector 2022 the only thing that needs to follow one down a page is the table of contents sandwich menu. And that is small enough that it could fit in the blank space on the left.

The sticky page title bar coexists with the sticky table headers on this German wikipedia page:

And it works with multi-row headers. And it has header cell borders.

Someone put a lot of work getting these sticky headers linked below to work. Look at relevant section in the CSS file.

--Timeshifter (talk) 05:53, 17 September 2023 (UTC)Reply[reply]

I have no idea what import style template is needed for, it seems overcomplicated for what it is trying to achieve, but this is a simple offset issue.
see also the gadget for sticky headers. MediaWiki:Gadget-StickyTableHeaders.css. unfortunately this causes other problems, when ppl start adding scrollable tables i side a scrollable page —TheDJ (talkcontribs) 14:24, 17 September 2023 (UTC)Reply[reply]
TheDJ and Tol. I am not talking about scrolling tables. To avoid confusion I changed the name of the relevant Help:Table section to this:
Help:Table#Tables with sticky headers. I left an anchor to the old section name since it has been used elsewhere.
The most recent previous discussion at this Village Pump:
Wikipedia:Village pump (technical)/Archive 206#Vertical scrollable table with sticky headers
And there is the Phabricator task you, TheDJ, started in 2012: Phab:T42763: "Implement repeated/fixed/floating/sticky table headers".
I don't want to wait years more to have something that works, even if it isn't perfect. I want something that works for everybody without installing a gadget. So a template is one way to go that could be used now.
{{Import style}} was created by Tol.
See Global Search at Toolforge. Search for
"{{Import style|sticky}}" - in quotes. To get transclusion count and list of articles. It currently is used on 93 pages on various-language Wikipedias. Those sticky table headers work fairly well in Vector 2010.
{{Import style}} is also used for something else. Maybe it can be split off into a single-purpose template with a descriptive name like:
{{Sticky headers}}. With class=sticky-headers
I would be happy with just sticky column headers for now. How about somebody do that and incorporate the German Wikipedia method, and other methods, into it, too? Mostly pinging people participating in, or mentioned, in past sticky table header discussions: Jroberson108. TheDJ. Tol. Newslinger. Sdkb. Graeme Bartlett. Bawolff. GhostInTheMachine. Yair rand. Izno. Jts1882.
When this Village Pump discussion is done, people can continue here:
Help talk:Table#Tables with sticky headers. Discussions. It links to past discussions too.
Are there forums just for templates?
--Timeshifter (talk) 12:40, 18 September 2023 (UTC)Reply[reply]
Template:Import style was my attempt at providing a single template with which multiple different reusable styles could be imported for use (to reduce redundancy between different style-importing templates), but it is not all that widely used. As for now, I can work on using the German Wikipedia implementation here (as it seems to work quite well there). Tol (talk | contribs) @ 13:55, 18 September 2023 (UTC)Reply[reply]
Sticky table headers seem to be the norm on all of my devices and I have yet to see a page where this seemed to be a problem. Then again, I do not normally use Vector2022. However, after a couple of tests with Vector2022, I am solidly confirmed in my resolve to stay with Real Vector, but the sticky table headers vs stick article header thing still does not seem to be a problem. As such, I don't see any need for a template. Is there a specific browser that seems to have this problem? — GhostInTheMachine talk to me 13:58, 19 September 2023 (UTC)Reply[reply]
True for Firefox, Edge, Chrome, and Opera. It is in Vector 2022 that the problems lie. {{Import style|sticky}} works well in Vector 2010. See: Help:Table#Tables with sticky headers for details. --Timeshifter (talk) 15:27, 19 September 2023 (UTC)Reply[reply]
@Timeshifter, am I missing something here? I don't see any issues with the sticky header on List of countries by real population density based on food growing capacity — Qwerfjkltalk 18:46, 19 September 2023 (UTC)Reply[reply]

For that list article you may need to narrow your browser width and/or increase the page's text size in order to better see the problems in Vector 2022. Here below is a simpler example: Part of a table from United States drug overdose death rates and totals over time#Comparisons to other countries. Note that when you scroll the page you can't see the sticky header in Vector 2022. But you can see it in Vector 2010.

Drug overdose deaths per year.
Location 2021 2020 2019 2018 2017
  Austria 235 191 196 184 154
  Belgium 168 152 148
  Bulgaria 20 24 11 24 18
  Croatia 77 99 97 85 65
  Cyprus 10 6 5 12 16
  Czech Republic 64 58 42 39 42
  Denmark 197 202 183 239

--Timeshifter (talk) 20:58, 19 September 2023 (UTC)Reply[reply]

You can also see it in MonoBook. --Redrose64 🌹 (talk) 20:54, 20 September 2023 (UTC)Reply[reply]

Redrose64. In Monobook skin are all the header lines sticky below in this table excerpt from United States drug overdose death rates and totals over time#1968–2021 overdose death rates and totals?:

1968–2020 US drug overdose deaths. 1,106,859 total.
Year Deaths Population
(July 1 residents)
Crude death rate per 100,000 Age adjusted death rate per 100,000
1968 5,033 199,533,564 2.5 2.8
1969 6,006 201,568,206 3.0 3.3
1970 7,101 203,458,035 3.5 3.8
1971 6,771 206,782,970 3.3 3.5
1972 6,622 209,237,411 3.2 3.4

--Timeshifter (talk) 14:58, 21 September 2023 (UTC)Reply[reply]

The row with the sorting arrows is non-sticky. The stuff above is all sticky. Sorry for the delay - catching up on my watchlist after four eleven-hour shifts and one six-hour. Not in that order. --Redrose64 🌹 (talk) 19:44, 22 September 2023 (UTC)Reply[reply]

Potential fix Edit

I have potentially fixed this by using:

body.vector-sticky-header-visible .is-sticky {
	top: 3.125rem;

This will only apply the additional 3.125rem vertical shift if the sticky header is visible at that time. Tol (talk | contribs) @ 22:13, 23 September 2023 (UTC)Reply[reply]

Thanks Tol! It works for me on Firefox, Edge, Chrome, and Opera.
Are there some simpler tables on German Wikipedia with sticky headers using the same template as the more complex table? And are they using a template above the table like here? Along with a class on the top line? I'd like to see how they are doing it. And how are they getting borders on headers?
I would prefer something simpler like {{sticky}} and class=sticky. --Timeshifter (talk) 00:13, 24 September 2023 (UTC)Reply[reply]
No problem, @Timeshifter. I'm not entirely sure how the German Wikipedia does this, but it looks like (in this instance) it's applied through a part of their de:Modul:Musikcharts system, which makes the header sticky (through styles at de:Vorlage:Charttabelle/styles.css) once the table reaches a certain length. My intent with Template:Import style was that multiple styles could be imported to a page through one system, in the hope of reducing duplication and redundancy, but it hasn't become too widely used. Tol (talk | contribs) @ 20:25, 24 September 2023 (UTC)Reply[reply]
Tol. I updated Help:Table#Tables with sticky headers to acknowledge that it is now working on Vector 2022 also. So, on German Wikipedia it looks like they don't have a simple dedicated template just for sticky headers? I think if such a template were created on English Wikipedia it would be widely used, and would be copied to other-language wikis. --Timeshifter (talk) 20:44, 24 September 2023 (UTC)Reply[reply]
Thanks for letting me know, @Timeshifter. I'm not aware of a widely used sticky-header template, though it looks like Template:Import style has been copied to some other language Wikipedias. Based on Wikidata (Template:Import style/sticky.css (Q110736505)), there are at least 5 other languages with Template:Import style/sticky.css. Tol (talk | contribs) @ 18:40, 26 September 2023 (UTC)Reply[reply]

Multi-row headers Edit

Tol. I noticed that only the top row of multi-row headers is sticky. That is a problem in some cases. It is interesting that only the top row loses the borders.

2018 Bauxite production and reserves (kilotonnes)
Country Production Reserves
 World 327,000 30,000,000
  Australia 110,000 6,000,000
  Guinea 82,000 7,400,000
  China 60,000 1,000,000
  Brazil 35,000 2,600,000
  Indonesia 23,000 1,200,000
  India 22,000 660,000

List of U.S. states and territories by incarceration
and correctional supervision rate
has several multi-row tables.

Adult incarceration rates by state. Dec 31, 2018
Jurisdiction 2018 rate per 100,000 of all ages 2018 rate per 100,000 adults 2018. Adults in prison or jail
  Federal 50 70 179,200
  State 580 750 1,919,200
  Louisiana * 980 1,280 45,700
  Oklahoma * 970 1,290 38,500
  Mississippi * 920 1,210 27,500
  Georgia * 850 1,110 89,700
  Kentucky * 840 1,080 37,500
  Alabama * 820 1,060 40,400
Correctional supervision rates by state, 2018.
Jurisdiction Total Community supervision Incarcerated
Total, 12/31/2018 Rate per 100,000 adults Probation or Parole, 12/31/2018 Rate per 100,000 adults In prison or jail, 12/31/2018 Rate per 100,000 adults
  Federal 302,100 90 122,800 40 179,200 50
  State 6,083,300 1,850 4,276,200 1,300 1,919,200 580
  Alabama * 99,200 2,030 60,900 1,240 40,400 820
  Alaska * 7,800 1,060 3,400 460 4,400 600
  Arizona * 136,800 1,890 84,300 1,170 54,600 750
  Arkansas * 74,700 2,470 53,800 1,780 24,700 820
  California * 513,100 1,290 312,400 790 200,700 510
  Colorado * 122,600 2,140 91,300 1,590 32,700 570

--Timeshifter (talk) 23:17, 24 September 2023 (UTC)Reply[reply]

@Timeshifter, yes, multiple lines of sticky headers are not going to work (and I'm pretty sure that they wouldn't have worked previously, either). Implementing an alternate method of styling to allow compatibility with multiple lines of sticky headers would probably not be compatible with the current implementation, but I can look into it to see if it would be feasible. Tol (talk | contribs) @ 17:29, 26 September 2023 (UTC)Reply[reply]
Tol. I tested both Vector 2022 and 2010 and neither work with multi-row headers.
class=floatable-header works with multi-row headers here: - it is a very large wiki.
The other year ranges linked at the top also have multi-row headers that float. It seems they are based solely on CSS: - search for "floatable".
I see nothing about floatable headers in the JS:
I am curious to know if the CSS can be copied as is to a template style of the same name:
{{floatable header}}
If I knew enough about templates and CSS I would try it myself. Maybe you or someone else could try it.
{{Floatable header}} and class=floatable-header are easy to remember.
Unlike {{Import style|sticky}} and class=is-sticky
That is why I say a dedicated template with a simple name would be used much more widely. I have difficulty remembering the current template here. It will never be widely used in my opinion on the English Wikipedia. Other wikis will try it out, but I doubt it will have wide use on the other-language wikis either. For the same reason.
I also like that class=floatable-header is used on the same line as class=wikitable. That is easier to remember too.
--Timeshifter (talk) 19:44, 26 September 2023 (UTC)Reply[reply]
@Timeshifter, taking a quick look at the Common.css you linked from the other wiki, it seems to be a collection of very messy workarounds (as acknowledged in the comments for the .floatable-header styles), and I would be very hesitant to use an implementation based on it. In particular, it is hard-coded to make each header row 29px below the previous one (resulting in it breaking quite badly when a header runs over multiple lines).
The reason that the floatable-header class is applied next to the wikitable class is because it is applied to the table, not to the row. This is because its CSS is done so that it applies to all header cells in the object (not to the object itself). Here, the CSS floats the object itself (the row).
I'm working on implementing an alternative style that would float the entire header (the thead element) of a table; this should be able to use the Vector 2022 header fix as well. Tol (talk | contribs) @ 21:58, 26 September 2023 (UTC)Reply[reply]
Just tested in Google Chrome (version 116.0.5845) with a Monobook skin, and scrolling seems to work perfectly: first, data rows slide up and hide beneath headers; then—when all data rows disappear—all header rows except the first one start hiding under the first one; finally only the first header row remains visible and it gets scrolled-out from view, too. --CiaPan (talk) 20:14, 26 September 2023 (UTC)Reply[reply]
CiaPan. I just tested in Chrome (version 117.0.5938.92) with a Monobook skin. The only header (in the multi-row header tables) I see floating is the top row. The rest roll under it as I scroll up. This is the same thing I see in the other browsers I tested: Firefox, Edge, Opera. --Timeshifter (talk) 20:28, 26 September 2023 (UTC)Reply[reply]

Yugipedia floatable-header tests Edit

Tol and all. See: - (Timeshifter was taken). I tested some tables there. This is the best I could do. See file history for what was done. That floatable-header CSS depends on scope=col

I couldn't get more header rows to float beyond what is showing there.

I created {{header test}} to try out some ideas. Feel free to edit it further. I haven't had time to try much yet in a sandbox. --Timeshifter (talk) 21:59, 26 September 2023 (UTC)Reply[reply]

@Timeshifter, as I mentioned above, the Yugipedia CSS works through a complicated and easily breakable workaround. The header scope parameters (scope="col") are there for accessibility reasons, and have their own meaning (see Mozilla docs); they really shouldn't be used for floating headers. Tol (talk | contribs) @ 22:03, 26 September 2023 (UTC)Reply[reply]

Multi-row headers implemented Edit

Timeshifter, I have put together a basic working implementation of multi-row headers. See:

An example table
A header above other headers
Header 1 Header 2
Another header 1 Another header 2
Content 1 Content 2
More content 1 More content 2
Even more content 1 Even more content 2

This can be done by applying the class is-sticky-head to the table itself; this then makes the table's entire header (thead element) sticky. There are still some problems with borders that I'm sorting out. Tol (talk | contribs) @ 22:13, 26 September 2023 (UTC)Reply[reply]

On second try, this seems to only work with sortable tables (as otherwise, there is no thead element, because MediaWiki doesn't implement HTML tables well). Tol (talk | contribs) @ 22:18, 26 September 2023 (UTC)Reply[reply]
Tol. It works in preview on all the tables in #Multi-row headers above. It does not work with {{Static row numbers table}}. Those subtemplates of {{Static row numbers}} are problematic and I don't use them anymore. I removed {{Static row numbers table}} from this thread. I substituted: class="wikitable sortable static-row-numbers srn-white-background" style=text-align:right;
--Timeshifter (talk) 23:29, 26 September 2023 (UTC)Reply[reply]

Tajik, optional istani Edit

A few days ago, a user edited the existing Category:21st-century Tajikistani women writers to change it from the manual coding to the {{Women writers by nationality and century category header}} template so that its formatting and coding would be automatically imported by the template — however, this had the effect of turning a bunch of bluelinked "Tajikistani X" categories into redlinked "Tajik X" categories.

Now, it was a good faith edit, I'm not accusing the editor of doing anything improper here — but we obviously can't leave it that way if the template is going to break stuff. But as happens more and more with templates of this type, the nationality labels that it chooses to transclude are not coded directly in that template, but are farmed out to some other module — meaning I can't find it in the first place, and even if I could find it I might be at risk of breaking other stuff in the process of trying to fix this broken stuff, so my only option for resolving the redlinks was to revert the edit and flip the category back to the old coding so that it went back into the Tajikistani categories that exist and wasn't left in Tajik categories that don't.

Accordingly, could somebody with more knowledge in the automated module area look into fixing this, so that we can avoid content being moved from Tajikistani bluelinks to Tajik redlinks in the future? Thanks. Bearcat (talk) 12:10, 20 September 2023 (UTC)Reply[reply]

Update: the same issue also resulted in a bunch of bluelinked "Luxembourgian X" categories turning into redlinked "Luxembourg X" categories at Category:21st-century Luxembourgian women writers. Bearcat (talk) 12:31, 20 September 2023 (UTC)Reply[reply]
I have never heard the word "Tajikistani" in any context--and I've been there. I've always heard the people called "Tajiks." It's not unprecedented that they're called "Tajiks" and not "Tajikistanis" or "Tajikis." The people of Afghanistan are called "Afghans" not "Afghanistanis." They're sometime erroneously called "Afghanis" which is the name of the the currency. Smallchief (talk) 14:27, 20 September 2023 (UTC)Reply[reply]

One can see "Tajikistani" used e.g. in our article Tajiks, section Tajiks#Pakistan. The word "Tajik" is originally a name of ethnicity, and "Tajikistan" is a state of Tajiks. As a result, "Tajikistani" could describe citizens of Tajikistan, who are not necessarily ethnic Tajiks. (Warning: WP:OR starts here! Such distinction may work, if an original nationality prevails in a state. However, it is not universal, for example, there is (AFAIK) no distinction between Portuguese or Ukrainians as nationality and Portuguese or Ukrainians as citizenship; similar distinction is sometimes not even possible, if a country (a state) is multi-national, like Brazil, South Africa or former Yugoslavia.) --CiaPan (talk) 15:53, 20 September 2023 (UTC)Reply[reply]

Perhaps the solution would be to ask a citizen of Tajikistan who is not a Tajik what they call themselves. I suspect it would not be Tajikistani. Tajikistani seems to me to only be used as an adjective not as a noun. Thus, in line with common usage and to avoid confusion, I changed the single use of the word "Tajikistani' in Tajiks#Pakistan to Tajik -- the word used throughout the article for the ethnic group. Smallchief (talk) 17:30, 20 September 2023 (UTC)Reply[reply]
  • I did not post here to spark a debate about what is or isn't the correct demonym for a person from Tajikistan — regardless of what name the categories should or shouldn't be at, the categories are where they are, so Tajikistan-related content has to use the categories that exist, and cannot be left sitting in redlinks that don't exist. Keep the category where it is or move it, that's none of my concern — but either way, whatever's done needs to be done across the board. Templates should not be autogenerating redlinked categories for one form that already exist as bluelinks at the other — so either the template has to be made to autogenerate the category that exists, or the category needs to be moved to the form that the template is going to autogenerate. A template autogenerating redlinks, because the category is actually at Form A while the template is coded to presume Form B instead of the form that the category is actually at, needs to never happen again. Bearcat (talk) 21:55, 20 September 2023 (UTC)Reply[reply]
    I'm simply pointing out that, to my knowledge, the word "Tajikistani" is used incorrectly and a category using that incorrect name should be renamed. You're looking for a fix. I can't help you there, but part of that fix is renaming the category. Perhaps: "Women writers of Tajikistan" and another category called "Tajik women writers." (not all Tajiks are citizens of Tajikistan and not all citizens of Tajikistan are Tajiks) Smallchief (talk) 15:40, 21 September 2023 (UTC)Reply[reply]
    I believe that Wikipedia has traditionally maintained a consensus that since you're correct that not all ethnic Tajiks live in Tajikistan, and not all citizens of Tajikistan are ethnically Tajik, it was necessary for the category tree to uphold a distinction between ethnic Tajiks and Tajikistani nationals. Regardless, if you want the category to be renamed, then feel free to list it for a CFR discussion — but you don't get to make it my job to do that: my job here was limited to being Destroyer of Redlinks. Bearcat (talk) 02:25, 24 September 2023 (UTC)Reply[reply]
Regardless of what's going on above, I've fixed the template so it doesn't autogenerate redlinks in this case via an edit to Module:Find demonym. * Pppery * it has begun... 01:17, 22 September 2023 (UTC)Reply[reply]

Dotted circle + combining diacritic displaying as tofu on Android web access Edit

The dotted circle ⟨◌⟩ is used heavily as a neutral grey background to demonstrate combining diacritics. This can be done either directly or via a template such as {{unichar}} using cwith= option. For example, combining acute accent can be given as ⟨◌́⟩ or U+0301 ◌́ COMBINING ACUTE ACCENT ({{unichar |0301 |combining acute accent |cwith=◌}}) Right now, the article diacritic is an illegible mess when viewed via Chrome on an Android phone (it works fine when viewed via Chrome on a Chromebook). Any ideas? 𝕁𝕄𝔽 (talk) 13:55, 21 September 2023 (UTC)Reply[reply]

Just to supplement: as seen from my mobile, the two freestanding dotted circles are displaying correctly but both cases where combining is used are tofu. 𝕁𝕄𝔽 (talk) 13:58, 21 September 2023 (UTC)Reply[reply]
I can confirm this, but it seems to me that Chrome on my Android phone just doesn't have access to / use the right fonts. —Kusma (talk) 20:14, 21 September 2023 (UTC)Reply[reply]
I don't see how that can be the cause because it certainly has the glyph for both the dotted circle and the combining acute (U+0301 ́ COMBINING ACUTE ACCENT). It is not a precomposed character. What is displayed is a tofu with a correctly placed acute accent. "It does not compute, Captain". --𝕁𝕄𝔽 (talk) 21:04, 21 September 2023 (UTC)Reply[reply]
Here is a precomposed o acute: ⟨ó⟩
and this is using cwith=o: U+0301 COMBINING ACUTE ACCENT. Which displays correctly! So what's wrong with the dotted circle which is the standard method 'frame' a diacritic. Curiouser and curiouser. --𝕁𝕄𝔽 (talk) 21:22, 21 September 2023 (UTC)Reply[reply]
Can confirm I'm seeing it as well (Chrome 117 on Android, fine elsewhere) - assuming that 'tofu' is the thing showing for me as [X]. However! Looking at diacritic, the examples in the lead are all broken, but further down the examples given for Hebrew niqqud etc are correctly showing the dotted circle. If we look at 'types', then they're mostly messed up but one or two are correct (interpunct, triangular colon, titlo) and for the combining ones, we get an [X] then the dotted circle for the second character. Mysterious indeed. Andrew Gray (talk) 22:24, 21 September 2023 (UTC)Reply[reply]
Yes, "tofu" is a nickname for .notdef, see unicode input#Glyph availability. But the crazy thing is that it does have the glyph. If it can manage to place the diacritic correctly on the tofu or on a "normal" letter, why can't it just place it correctly on a dotted circle?
In the cases where the dotted circle is shown, the rendering is not actually being done correctly. See for example diacritic#Greek: the iota subscript is appearing after the circle rather than under it. I don't speak Hebrew but I suspect the same is happening with niqqud too. 𝕁𝕄𝔽 (talk) 08:23, 22 September 2023 (UTC)Reply[reply]
Overall this sounds like something that should be reported upstream somewhere (although where exactly I don't know). A minimal reproduction seems to be an HTML file containing something like <span style="font-family: sans-serif">◌͝◌</span>, no need to bring Wikipedia or MediaWiki into it. Anomie 12:07, 22 September 2023 (UTC)Reply[reply]
A simple acute accent (<span style="font-family: sans-serif">◌́</span>) would be even simpler. I guess unless someone has a back channel to Google to make a bug report, we just have to accept it as a client end browser deficiency and so not something that we can fix here. A fix for your "two character spanning diacritic" would be icing on the cake.
For now, diacritic begins with {{Contains special characters}} and I will add (precomposed) examples where I can, in the cases where the rendering has failed.
I think we have taken this one as far as we can here, my thanks to all who contributed. --𝕁𝕄𝔽 (talk) 12:25, 22 September 2023 (UTC)Reply[reply]
I see this problem on my Chrome for Android. I opened up remote debugging to check which font-families are being chosen. On my phone's Chrome, the correctly displaying dotted circle on its own displays in SEC Naskh Arabic (bizarre font choice), and the tofu dotted circle with non-tofu combining acute displays in Roboto. On my phone's Firefox, the dotted circle and the dotted circle with combining acute both display in Noto Naskh Arabic. In both browsers, the angle brackets are in Roboto. So Chrome is choosing to display the grapheme cluster in a font that contains a glyph for one code point but not the other, whereas Firefox chooses a font that works for both. — Eru·tuon 19:40, 22 September 2023 (UTC)Reply[reply]
Thank you, that explains well what is happening. As to why it is happening, God (aka Google) only knows. And to think that the Noto (No tofu) fonts were supposed to be the magic bullet for such issues! --𝕁𝕄𝔽 (talk) 20:35, 22 September 2023 (UTC)Reply[reply]
Roboto is the older font family that predates Noto. Some devices will use Noto, but many stylesheets still call for Arial. Andre🚐 20:44, 22 September 2023 (UTC)Reply[reply]
Result! That was the required hint. This syntax fixes the problem: <span style="font-family: Noto, Roboto, serif, sans-serif">◌́</span> produces ◌́. Checked on android phone and Chromebook, both using Chrome. Of course it probably wouldn't work on iPhones or Windows PCs now, but who cares?   But to be serious, would someone please suggest suitable fonts to add to the font family string, to keep other vendors on board. The fun part will be revising Unichar accordingly (come back, DePiep, all is forgiven?). — Preceding unsigned comment added by JMF (talkcontribs) 23:14, 22 September 2023 (UTC)Reply[reply]
If we're going to mess around with the font string generated by the template, just <span style="font-family: serif"> was sufficient in the quick testing I did. OTOH, having random serif font in other places (like "áéíóú◌́" rather than "áéíóú◌́") might look odd or cause other problems. Anomie 23:47, 22 September 2023 (UTC)Reply[reply]
The problem is just because some devices have their default sans-serif font set to Roboto. It can be fixed by giving any font other than sans-serif. You wouldn't want serif because then you'll get serifs. I happen to have Noto Sans installed, so everything looked fine for me, but you could do Noto, sans-serif, which would display Noto to anyone who has Noto, and fallback to Roboto, or Arial or whatever font the device has set for sans-serif. Andre🚐 01:31, 23 September 2023 (UTC)Reply[reply]
I had already found the simple serif trick but I didn't have consistent results. But let's assume it will work. I don't speak templatish so this may be a hand-waving answer, but... The problem only arises in two cases:
  1. in articles about diacritics, where the ◌ is being used as "neutral grey background". I see no difficulty here, indeed quite often serif is needed to properly distinguish the glyph. It will mean a series of manual updates but I am happy to do that.
  2. {{unichar}} when cwith=◌ is used. This is again almost always to illustrate a diacritic. The 'fun' will be to revise that complex template to have a display that is conditional on the presence or otherwise of cwith=◌. But let me take that one back to template talk:unichar to see what might be involved.
Anything more to say? --𝕁𝕄𝔽 (talk) 08:03, 23 September 2023 (UTC)Reply[reply]

If there is a kind template wizard in the house, a small enhancement to {{unichar}} is needed, to handle cwith=◌ with a bit more style . I have stated the requirement at template talk:Unichar#emplate enhancement needed, please. --𝕁𝕄𝔽 (talk) 16:45, 23 September 2023 (UTC)Reply[reply]

I have extended {{script}} which can be used as the use= option for unichar, to support use2 as {{script/Noto}} or {{script/Serif}} (pass as lowercase to the param) e.g.
  • {{unichar |0301 |combining acute accent |cwith=◌|use=script|use2=serif}}
  • {{unichar |0301 |combining acute accent |cwith=◌|use=script|use2=noto}}
I have applied the above change to use "serif" whenever cwith=◌ is used. See if it works. Andre🚐 00:25, 24 September 2023 (UTC)Reply[reply]
Again, thank you for your help. The change to {{unichar}} is live and working well. I have begun the longwinded process of updating the individual diacritic articles that use the "dotted circle + combining diacritic" directly.
This discussion may now be regarded as concluded satisfactorily. My thanks to all who contributed, especially Andrevan. --𝕁𝕄𝔽 (talk) 15:09, 24 September 2023 (UTC)Reply[reply]
In which case,
. --Redrose64 🌹 (talk) 19:35, 24 September 2023 (UTC)Reply[reply]

Replag or Thursday issues? Edit

Hello, VPT,

I was wondering if there are system problems today. I'm finding broken redirects that aren't showing up on the bot report that reports them and other pages that should show up on Quarry queries that are coming back as empty. Is there a system problem going on today that is preventing information from being updated? Thanks for anyone who knows enough to know whether to check if there are open tickets that could explain what's happening. Liz Read! Talk! 19:58, 21 September 2023 (UTC)Reply[reply]

I forgot to mention that XTools is also not updating counts. Not sure if this is related. Liz Read! Talk! 20:27, 21 September 2023 (UTC)Reply[reply]
There's a little bit of replication lag currently, but I'm not sure if that's causing the issues you describe — could you link to an example? — TheresNoTime (talk • they/them) 20:35, 21 September 2023 (UTC)Reply[reply]
A little bit of replication lag? It's up to three hours now. Is replication broken, or is it just on pause for a good reason? (I think I recall reading somewhere that schema updates cause replication to be stopped for a while, right?) --rchard2scout (talk) 22:31, 21 September 2023 (UTC)Reply[reply]
Well, I noticed in the examples I mentioned. The Quarry queries I run are coming up empty. For example, in this one, User:Monique_Brynnel ia indicated as a nonexistent user page but that page was deleted hours ago and no longer exists so it shouldn't show up as existing. Another Quarry query and AnomieBOT III's report shown 0 Broken redirects when I can easily find some broken redirects when looking at the Move logs. But most obviously, I've done a lot of page deletions today and according to XTools, I've only deleted 10 pages today when it's at least over 100 pages. My number hasn't increased for the past 12 hours. That's where I have noticed the anomalies. Liz Read! Talk! 23:33, 21 September 2023 (UTC)Reply[reply]
The most recent revision on the replicas is currently Special:Diff/1176409855, a bit under nine hours old and twice what is showing. I suspect phab:T343198 is to blame. —Cryptic 00:07, 22 September 2023 (UTC)Reply[reply]
Thank you, Cryptic, as long as someone competent is looking into this. Also, Wikipedia:Database reports/Empty categories is showing no new empty categories all day today except those already at CFD which is highly unusual. I guess at some point in the next 24 hours, things will catch up. Liz Read! Talk! 01:11, 22 September 2023 (UTC)Reply[reply]
I don't see the connection between T343198's issue and bot and Quarry reports and XTools but I'm no technical wizard. Liz Read! Talk! 01:15, 22 September 2023 (UTC)Reply[reply]
It's a database schema change to add a column, which, as noted above, may well cause issues like this. Graham87 (talk) 03:04, 22 September 2023 (UTC)Reply[reply]
Graham87, any idea when this problem will be resolved? None of the bot reports I rely upon are up-to-date today. The "special pages" are fine but all of the database reports are not updating. I can't believe that no one else has come to this board to comment upon these delays. Liz Read! Talk! 05:31, 22 September 2023 (UTC)Reply[reply]
No idea. Graham87 (talk) 06:22, 22 September 2023 (UTC)Reply[reply]
Currently caught up. There's still some maintenance to be done on s1 (which is where we are), but I am very much not "someone competent" for this purpose; I can't begin to predict the impact, if any, the part that's left will have. —Cryptic 23:19, 22 September 2023 (UTC)Reply[reply]

Articles with photos in category Edit

Is there an easy way to identify articles which contains photos in a particular category on Commons, other than going through that category manually? Nikkimaria (talk) 01:35, 22 September 2023 (UTC)Reply[reply]

@Nikkimaria The glamorous tool can do this (example) - it's intended for crosswiki use and it's not possible to filter it to just enWP, but the list does break down which project & pages each file is used on ("file usage details" tab). Andrew Gray (talk) 12:04, 22 September 2023 (UTC)Reply[reply]

Link Previews are failing Edit

[iOS, Mobile Web]. In Safari, there is a feature where, if you continuously touch a link, a preview of the webpage will be downloaded and display. On English Wikipedia, however, the preview will display the same page the link is in. All other language Wikipedias, and other English Wikimedia sites, and the dektop version of WP accessed in Safari do not have this glitch Mach61 (talk) 01:22, 23 September 2023 (UTC)Reply[reply]

Example of such a link? Nardog (talk) 02:34, 23 September 2023 (UTC)Reply[reply]
All wikilinks, regardless of namespace - but not external links Mach61 (talk) 04:24, 23 September 2023 (UTC)Reply[reply]
have you tried this logged out ? Because the most likely cause is a userscript or gadget. —TheDJ (talkcontribs) 08:43, 23 September 2023 (UTC)Reply[reply]
@Mach61 I assumed you are refer to this iOS feature. I have iPhone (iOS 17) and is working for me on Safari. As this is iOS native feature, I don't see how it's related to Wikipedia. Considered trying on other website to see if you are able to link preview. If you are not referring to the iOS feature, then kindly provide the detailed steps to replicate. Paper9oll (🔔📝) 08:54, 23 September 2023 (UTC)Reply[reply]
I am talking about that specific feature, and all other sites work with it, other than “”. Again, both the desktop skin and other language mobile versions of Wikipedia work. Mach61 (talk) 13:41, 23 September 2023 (UTC)Reply[reply]
@Mach61 My last available solution is try appending ?safemode=1 to the end of the URL on the page you're viewing and hit enter, after page refresh, try touch and hold any wikilink to see if it works. If this works then it's your installed userscripts and/or gadgets blocking such behaviour which to me personally is unlikely/unexpected scenario since as far as I'm aware of, website cannot manipulate native iOS feature such as the Safari's link preview. Also check Special:Preferences#mw-prefsection-rendering, enable page previews is ticked. Paper9oll (🔔📝) 16:09, 23 September 2023 (UTC)Reply[reply]
Appending “?safemode=1” doesn’t work. Will inquire into other solutions. Mach61 (talk) 17:17, 23 September 2023 (UTC)Reply[reply]
Please post an actual example instead of just claiming it's on all pages. You obviously haven't tested millions of pages. Always post an example if possible when you report a problem. It works correctly for me in iOS 16.6.1 on an iPhone 8, for example on Mach61 in your signature on this page. Does that fail for you? Does the link go to your user page when you tap it normally? PrimeHunter (talk) 18:43, 23 September 2023 (UTC)Reply[reply]
No, clicking that link doesn’t preview my page; it previews VPT. Yes, links work normally. I have an iPhone 12 on IOS 17.0.1 Mach61 (talk) 01:24, 24 September 2023 (UTC)Reply[reply]
Update: a few minutes ago, I was directed to my Wikimedia central login via preview, with a message that my token was invalid. After clicking “dismiss” on that page, the bug still exists as before Mach61 (talk) 05:37, 24 September 2023 (UTC)Reply[reply]
Second update: at Wikipedia:Bureaucrats' noticeboard/Archive 49#, this link to a special page previews correctly. Other wikilinks still do not. Mach61 (talk) 02:48, 26 September 2023 (UTC)Reply[reply]
I think I’ve narrowed down the issue to having something to do with the preview working with the last page in my history/in my cache/????. This is because when I click Special:Random, previews for those links will also be for a different random page. Again, “” is the only website this occurs with Mach61 (talk) 02:59, 26 September 2023 (UTC)Reply[reply]

Extention to go directly to changed content from the watchlist? Edit

While the watchlist can link to a specific section on a page, it can't go straight to the actual content that is changed. I can guess why this is the case— sections have anchors and arbitrary text does not— but I wonder if it's possible to hack something up like a) detect the first instance of changed content from the diff and b) insert a temporary anchor into the watchlist target and link directly to that. Or whatever. I frequently wish for this when people add comments to long, deep talk page conversations, where the diff doesn't allow me to naturally place the new comment in proper context. And yes, there are edge cases like multiple instances of changed content, but for my aforementioned scenario, going straight to the first instance of changed content would suffice. Does something like this exist, or could it ever? Orange Suede Sofa (talk) 22:43, 23 September 2023 (UTC)Reply[reply]

For signed posts, there is already an anchor, it's crucial to the reply feature - for instance this post. --Redrose64 🌹 (talk) 23:30, 23 September 2023 (UTC)Reply[reply]
Ah yes, there it is. Now to only get the watchlist to know about it and link accordingly... Orange Suede Sofa (talk) 00:18, 24 September 2023 (UTC)Reply[reply]
Not the same but similar: is there a way to go directly to the diff of unviewed changes from the watchlist? Or anything that can shorted the chain of watchlist -> history -> select last revision that doesn't say "updated since your last visit" -> click "compare selected revisions"? LittlePuppers (talk) 02:16, 24 September 2023 (UTC)Reply[reply]
The grouping mode has this (Preferences → Recent changes →   Group changes by page in recent changes and watchlist). Nardog (talk) 02:27, 24 September 2023 (UTC)Reply[reply]
Ah, excellent! I never thought about this as a consequence of that setting, but it makes sense. (I might see about doing something so the list of edit summaries is expanded by default, but I'll try it like this first.) LittlePuppers (talk) 02:33, 24 September 2023 (UTC)Reply[reply]
Oh, that only collapses edits that occur consecutively, without edits to other pages on my watchlist happening in the interim. LittlePuppers (talk) 04:04, 24 September 2023 (UTC)Reply[reply]
If you want to see all the changes since your last visit, you can click on the "cur" link at the start of the history item corresponding to the last revision you've seen already, thus skipping selecting the radio button and clicking on "Compare selected revisions". isaacl (talk) 03:14, 24 September 2023 (UTC)Reply[reply]
Good suggestion, that makes sense. LittlePuppers (talk) 04:04, 24 September 2023 (UTC)Reply[reply]
It's not exactly what you ask for, but I found that visual diffs are usually much nicer for reading new comments on talk pages than the wikitext diffs. You can view them by using the "Visual / Wikitext" toggle buttons in the top-right corner when viewing a diff. Matma Rex talk 15:39, 25 September 2023 (UTC)Reply[reply]

How to make a new template Edit

How does wone go about making a new template for a sport. I am trying to make one for netball tours. I have though no idea where to begin. PicturePerfect666 (talk) 04:41, 24 September 2023 (UTC)Reply[reply]

See Help:Templates#Writing templates Mach61 (talk) 05:32, 24 September 2023 (UTC)Reply[reply]
What exactly do you want the template to show? There are lots of templates that show the results of matches or tables of results. There is probably some existing template that does approximately what you want and you can use that as the template for your template. —  Jts1882 | talk  08:40, 24 September 2023 (UTC)Reply[reply]
Well in the same way cricket has a cricket series template I would like to generate a similar template for netball. cricket and rugby have tours, and so does netball and I am trying to create a template in the same vein for those tours. There is no existing template which does what I want. PicturePerfect666 (talk) 14:33, 24 September 2023 (UTC)Reply[reply]

Is there a way of getting an individual to help guide me through how to create a template? As I am struggling, Thank you in advance. PicturePerfect666 (talk) 16:12, 26 September 2023 (UTC)Reply[reply]

Reflinks not working? Edit

None of the 69. links at WP:REFLINKS seem to be working. ♦IanMacM♦ (talk to me) 08:57, 24 September 2023 (UTC)Reply[reply]

Still not working. Is there any way to fix this, as User:Dispenser is not active nowadays.--♦IanMacM♦ (talk to me) 14:39, 26 September 2023 (UTC)Reply[reply]
@Ianmacm, there is not. The code is not freely licensed so cannot be reused. — Qwerfjkltalk 15:39, 26 September 2023 (UTC)Reply[reply]
In that case I'm officially pissed off, because it's one of my favourite tools.☹️--♦IanMacM♦ (talk to me) 16:09, 26 September 2023 (UTC)Reply[reply]
@Ianmacm: See the archives of this page, going back to mid/late 2013, for a number of threads concerning Dispenser's tools, why they are not hosted on WMF servers, and therefore why they cannot be maintained by anybody else.
Basically: Dispenser's tools were hosted on Toolserver, the hardware for which was owned by Wikimedia Deutschland and who took it down at the end of June 2014. The WMF provided Tool Labs as a replacement, but required that anything on there had to be freely-licensed and, I believe, open source as well, conditions that Dispenser refused to comply with. So Dispenser hosted their tools elsewhere, on hardware that has since become non-functional. Hence this edit, amongst others. --Redrose64 🌹 (talk) 21:33, 26 September 2023 (UTC)Reply[reply]
Thanks. I've edited WP:REFLINKS to make clear that WP:REFILL is now the only similar tool that is available.--♦IanMacM♦ (talk to me) 07:47, 27 September 2023 (UTC)Reply[reply]
Providing supported replacements for Dispenser's invaluable tools is a perennial request at the annual Community Wishlist Survey, but would require a lot of work and has so far had no resource allocated. Certes (talk) 10:43, 27 September 2023 (UTC)Reply[reply]

Template:Cite Gnis2 problem Edit

The second {{Cite gnis2}} template on Vindication Island links to when it should link to Specifying |type=antarid causes it to point to an archived webpage, instead of the correct URL. Checking Template:GNIS URL which controls the URL it seems like there is no way to link through these templates; can it be changed so that it does? Or at least point to the correct template? Jo-Jo Eumerus (talk) 09:05, 24 September 2023 (UTC)Reply[reply]

Please unfuck page history Edit

The date picker in page histories sucks for two reasons.

  1. One has to enter a full date YYYY-MM-DD. Not even entering a year and clicking on 1 January works.
  2. After submitting a date, one cannot browse to edits that are more recent than the date selected.

The following gadget does not need a full date and allows browsing in both directions. One has to be on a page history in order for it to work. The date string may be one of the following:

javascript:var pleaseunfuckpagehistory = new URL(location.href); pleaseunfuckpagehistory.searchParams.set('offset', prompt('Please enter date string','')); location.href = pleaseunfuckpagehistory;

I hope this will be useful to you. Utfor (talk) 19:36, 24 September 2023 (UTC)Reply[reply]

Yeah, I edit the url manually instead of fighting with the date widget, too. —Cryptic 01:05, 25 September 2023 (UTC)Reply[reply]
"Not even entering a year and clicking on 1 January works." – At least this seems easily fixable. Clicking on any other date actually works, it's just the first day of the year/month that confuses the widget. I filed T347319 and proposed a patch for it. Matma Rex talk 16:26, 25 September 2023 (UTC)Reply[reply]

Is there a way to get all STEM pages on wikipedia? Edit

Question - Imagine that I am a STEM educator and wanted to get a complete list of all on-topic categories, subcats, and page ids for just STEM articles on wikipedia. How would I do that?

My first attempt was looking at the categorylinks table - mw:Manual:Categorylinks_table, but anyone familiar with that would know that it's a dead end. Wikipedia is too over categorized (eg, "coat of arms with plants" is under botany->plants

I've also asked this in other places:

Wikipedia:Village_pump_(policy)#Category_views; d:Wikidata:Project_chat#Is_there_a_way_to_get_all_STEM_pages_on_wikipedia?; Wikipedia_talk:WikiProject_Council#Get_a_complete_list_of_STEM_categories,_their_subcategories,_and_page_ids Wikiqrdl (talk) 00:33, 25 September 2023 (UTC)Reply[reply]

Since geology categories end up including continents which end up including countries and all things and people in those countries, category descent is not the way to go. You could attempt WikiProjects. EG gives all the chemicals project tagged pages. Graeme Bartlett (talk) 01:12, 25 September 2023 (UTC)Reply[reply]
@Graeme Bartlett yeah, exactly. I've poked the wikiprojects people, though they are more about people than data. It's some combination of wikiprojects and portals, I think. I've pinged the dbpedia people as well, as I believe they were trying to solve this problem. Wikiqrdl (talk) 01:25, 25 September 2023 (UTC)Reply[reply]
I would use the templatelinkstable. Most pages under a wikiproject have an wikiproject template. (remove the colon in sql) The pages in the sql query below are talk pages, so not the page_id you are looking for. Since you did look at the categorytable, I suppose you can try to figure out to couple it with the sql, and find the correct page_id. After all, people do not learn by being fed every single detail.
select page_title from page
join templatelinks on tl_from = page_id
join linktarget on tl_target_id = lt_id and lt_title in ("maths_rating", "WikiProject_Technology", "WikiProject_Science", "WikiProject_Engineering")
where page_namespace = 1
and page_is_redirect = 0
--Snævar (talk) 02:27, 25 September 2023 (UTC)Reply[reply]
ahh, I missed that. They're tagging the talk pages. Fascinating! Thanks. Wikiqrdl (talk) 05:29, 25 September 2023 (UTC)Reply[reply]
Not all relevant articles would necessarily be tagged, so there is probably value in using ORES topic predictions in addition to the wikiproject tagging. Now, running the model yourself on every article would be a massive inconvenience, but as it turns out, this is built into the default search function (as documented at mw:Help:CirrusSearch#Articletopic). Indeed, Special:Search/articletopic:stem returns slightly under 1.16 of the 6.72 million articles on the english wikipedia, and at a quick glance most results seem relevant. The best way to export that, I'm not yet sure, but it will probably be through the API. Alpha3031 (tc) 11:28, 25 September 2023 (UTC)Reply[reply]
In the end this is a lack of scope problem. There is no hard line where 'STEM' stops, so in a system like Wikipedia that means everything will bleed into your result set. —TheDJ (talkcontribs) 11:36, 25 September 2023 (UTC)Reply[reply]
PetScan may also be worth a look (though it appears to be down temporarily). It can limit to pages with any or all of certain templates, and there's an option to check the talk page rather than the article. Certes (talk) 13:01, 25 September 2023 (UTC)Reply[reply]
I've used PetScan + templates at scale before - it is a bit fiddly sometimes and prone to falling over, but it's an amazing tool for this and absolutely second the recommendation.
For something conceptually fuzzy like "STEM pages" inferring it from widely-used templates is probably going to be the best approach possible, short of text-classifying tools - if those templates exist. (I did an analysis on # of sports bios a while back that lucked out by being able to use article infoboxes, for example). You'll miss a lot of stuff, and get a lot of false positives, but it'll still be reasonably OK for a high-level approximation.
The category tree is hopelessly muddled and Wikidata is probably far too fine-grained and complex for this. Andrew Gray (talk) 21:28, 27 September 2023 (UTC)Reply[reply]

Pages using WikiHiero category rename Edit

Related to the above Wikipedia:Village pump (technical)#Pages using WikiHiero discussion, yesterday JJMC89bot was set loose on speedy renaming the category to Category:Pages using the WikiHiero extension — however, as that's a template-transcluded category, it did not have the result of moving any of the pages into the new category, so that the new category was empty and there were 546 pages left sitting in the redlinked old category, which was (obviously) thus sitting on the latest run of Special:WantedCategories.

Per past discussion on similar cases, however, JJMC89bot is not supposed to move categories that are template-transcluded if it can't make the necessary template edits to automatically move the contents over to the new category, and moves of that type have to be performed manually so that they don't create followup chaos like this.

But since I can't figure out what template to edit so that the pages move to the correct category, I've had to temporarily revert that category renaming so that it can be handled correctly by a human. Can somebody with more knowledge of that situation ensure that whatever template was transcluding the old category gets properly updated so that the category can be moved back to the updated name again? Thanks. Bearcat (talk) 12:44, 25 September 2023 (UTC)Reply[reply]

I submitted an edit request at MediaWiki talk:Wikhiero-usage-tracking-category and listed the category at Wikipedia:Categories for discussion/Working/Manual (although following the instructions led to unusual formatting). This was a speedy renaming request processed by Ymblanter, who maybe missed that it was not a normal category renaming for the bot. I don't usually get involved with category renames, so I'm stumbling in the dark a bit. – Jonesey95 (talk) 12:57, 25 September 2023 (UTC)Reply[reply]
We can not decline requests (speedy or not) just on the basis that the category is transcluded by template, and someone has to solve these cases afterwards (we have another bunch just hanging out there). Declining requests would amount to overwriting consensus. To be honest, I am already prepared to remove all these templates and add categories by hand. It might have been a brilliant idea, but it created more trouble what it is worth. Ymblanter (talk) 13:56, 25 September 2023 (UTC)Reply[reply]
I don't think anybody is declining the request. The problem appears to be that the category renaming was assigned to a bot when it should have been handled manually. – Jonesey95 (talk) 15:26, 25 September 2023 (UTC)Reply[reply]
Could you please point out to the discussion where such consensus was established? Ymblanter (talk) 13:57, 25 September 2023 (UTC)Reply[reply]
The category renaming was proposed as a speedy rename because the name assigned by the WMF developers is suboptimal (I'm trying to be nice). There were no objections. – Jonesey95 (talk) 15:29, 25 September 2023 (UTC)Reply[reply]
Thanks, but I was asking about a general consensus to not feed the template populated categories to a bot. Ymblanter (talk) 16:18, 25 September 2023 (UTC)Reply[reply]
I don't know of a discussion, but I would not recommend feeding such category renames to a bot on the grounds that it generally won't work. I don't think you need consensus to avoid doing something that doesn't work. Maybe I misunderstand. – Jonesey95 (talk) 16:27, 25 September 2023 (UTC)Reply[reply]
Well, checking if every category is added manually or by a template increases the length of my workflow by a factor of three. And for the last month, I have been the only administrator dealing with speedy CFD. For the last few years, we were feeding such categories to a bot which would create the new category, and then someone (typically me) would fix the template so that the articles get to the new category. But there have been several cases recently when the template was too complicated. I do not think the solution is to force me to check every single category, especially if there is no consensus it is strictly necessary. Ymblanter (talk) 16:40, 25 September 2023 (UTC)Reply[reply]
Except that if you don't check, you increase the length of everybody else's workflow by a factor of three thousand, because it causes fusterclucks like this where a bot moves the category, but fails to move any of the category's contents over to the new category, so the new category is left empty while all of the contents are left malingering in a redlink. In other words, what you're saying is that your time is more valuable than everybody else's: you don't want to spend a few extra minutes to check if a category is being template-generated before throwing it to a bot, but you don't care about the couple of hours that everybody else has to put in to fix a problem that could have been prevented with a simple two-minute check? It's not terribly helpful to make messes that other people have to invest even more time into cleaninug up, you know.
So if you don't want to invest more time into speedy renamings than you already do, and I deserve to be able to invest a lot less time into redlinked category cleanup than I currently have to, then how about a compromise? Maybe, when you're speedy renaming categories, how about the bot doesn't delete the old category right away, and instead leaves it in place as a category redirect, which can then be revisited and deleted once it's been confirmed as emptied? Maybe it could even have its own special "categoryredirect" template different from the regular one, which would function like dated maintenance categories and automatically add the old category to the speedy deletion queue once it was emptied? That way the bot can freely move the category and thus save you some valuable time, but the old category doesn't become a redlink until after it's been depopulated so that my valuable time is also saved? Does that sound like a fair middle ground? Bearcat (talk) 19:29, 25 September 2023 (UTC)Reply[reply]
Perhaps the Speedy CFD instructions could be enhanced to emphasize that adding "NO BOTS" to the beginning of a CFD nomination greatly improves the odds that everyone involved will do the right thing. Pppery did make a note in their nomination that a MediaWiki file would need to be edited, but that note apparently was missed by Ymblanter when they heroically processed the nomination. Gnomes can be thin on the ground sometimes, so I get it; some days, it feels like I'm the only one who cares about some essential chore. In this case, we could all decide to be on the same side against hasty work by the developers who made this work for us in the first place, but even better would be for us all to pull together. – Jonesey95 (talk) 21:39, 25 September 2023 (UTC)Reply[reply]
Did you mean hieroically?
Trappist the monk (talk) 21:44, 25 September 2023 (UTC)Reply[reply]
It's a fair cop. I can't believe I missed that golden opportunity, especially given the missing "i" at MediaWiki:Wikhiero-usage-tracking-category! – Jonesey95 (talk) 21:51, 25 September 2023 (UTC)Reply[reply]
Yeah, and as one of CFDW's two regular admins (Ymblanter being the other one - echoing Jonesey95's point about gnome tasks often being understaffed), I should have known that and I see the instructions do say to use NO BOTS. But I must have thought that pointing out the exact edit page that needed to be edited to carry out the proposal was sufficient. * Pppery * it has begun... 01:17, 26 September 2023 (UTC)Reply[reply]
The botop thinks it's the listing's admin's responsibility not to feed the bot bad instructions (or at least did so the previous time you complained). * Pppery * it has begun... 01:17, 26 September 2023 (UTC)Reply[reply]
Ok, no problem, I will start checking every category I feed. But then nobody should be surprised the speed goes down factor of three. I only have so much time to invest every day. Every admin is of course welcome to help. Ymblanter (talk) 05:07, 26 September 2023 (UTC)Reply[reply]
All over Wikipedia, but especially at CfD (as well as other places), there is a lack of admins either willing or competent to help. — Qwerfjkltalk 15:40, 26 September 2023 (UTC)Reply[reply]
If a redirect is an acceptable temporary solution I can easily do it manually. But I am afraid on top of it we have empty new categories which then get deleted within seven days. Again, I am more and more thinking about just direct ban on the categories populated by templates. Ymblanter (talk) 05:09, 26 September 2023 (UTC)Reply[reply]
(side note) Huh, these category names sure are a mess. I filed a bug to change the default ones to match the ones used here on English Wikipedia: T347324. Maybe someone is bored enough to work on this :) Matma Rex talk 16:48, 25 September 2023 (UTC)Reply[reply]

Module Routemap Edit

Another redlink on the latest run of Special:WantedCategories for Category:Pages with errors of Module Routemap; both of the pages in it relate to a coding error in the same {{Main Western Line}} routemap. In both articles, the routemap is specifically displaying a "Missing start-Collapsible markup!" line near the bottom of the map; however, since I'm not knowledgeable about that module's coding, I don't know how to fix it myself, so I wanted to ask if somebody with more expertise can fix that.

But also, even if and when that error is fixed and the category becomes depopulated, I note that virtually all of the other "Pages with errors" maintenance categories that we do have are named errors in template-or-module, rather than errors of template-or-module. So even though this seems like a maintenance category that probably should exist to catch future errors in routemap usages, I don't want to create it at "of" if "in" is more standard. So if the category is desirable, then could somebody also ensure that the module generates it as "in" instead of "of", and then create it accordingly? Thanks. Bearcat (talk) 13:20, 25 September 2023 (UTC)Reply[reply]

  Fixed. – Jonesey95 (talk) 16:22, 25 September 2023 (UTC)Reply[reply]

Template:Coord at the top breaks page previews Edit

See, for example, Niš. This would be fixed by removing {{Coord}} output from the parser by adding noexcerpt class, see the documentation for TextExtracts. stjn 19:45, 25 September 2023 (UTC)Reply[reply]

@Stjn, Module_talk:Coordinates#Protected_edit_request_on_29_May_2023. You may wish to comment. Izno (talk) 20:47, 25 September 2023 (UTC)Reply[reply]

Reclaiming my volunteer status Edit

Hello, all,

At the end of this week, I will be reclaiming my volunteer status.

I joined the Wikimedia Foundation in 2013 because I believed that the introduction of the visual editor would significantly affect here at "my home on the internet".  I wanted to be able to help the community through the inevitable bumps on the road, and to help the Editing team understand what experienced editors need.  That path took longer than any of us had expected, and some parts of it were bumpier than any of us wanted, but the Editing team finished the deployment process last month.  From the day it arrived here at the English Wikipedia on July 1, 2013 until when it finally got out of Beta Features at the very last wikis took 10 years and 53 days. Combined with the mobile visual editor and the 2017 wikitext editor, more than 120 million edits have used VisualEditor, including more than 27 million here at the English Wikipedia.

We have been down a few other paths together during that time (remember Typography refresh in 2014, or Special:LintErrors in 2016?), and as I think about transitioning back to volunteer life, I have two projects that I want you all here at VPT to keep in mind during the coming months.  

The first is DiscussionTools, which was the main result of the massive Talk pages consultation in 2019.  The last bits in active development are nearly finished, and the last piece will probably be deployed in the coming months.  It will add information to Talk: pages (e.g., the number of comments in a section), and therefore it will change the appearance of the talk pages.  It seems to be pretty popular and effective, so I don't expect you to get too many questions about it, but the key thing for folks on this page to know about this is:  If you don't like it, you can turn it off in preferencesIn fact, if you wanted to avoid it entirely, you could go to Beta Features now, enable Discussion tools, save your new prefs setting, and then go to the Editing prefs and turn it off (it'll be very last item in the list, "Show discussion activity"), save that change, and then you'd never even see it when it's deployed.  You might, though, want to first give it a try for a few days, to see whether you like it; I do.

The other is that some time next year, all the IP addresses will be hidden, e.g., in page histories.  Don't worry; almost everyone reading this page will still be able to see the IP addresses for unregistered editors; it's just the new editors and the general public who won't have access.  The team working on the new Temporary accounts system has been talking to the stewards about patrolling needs this year, and they're working on a new-and-improved version of, which should help ease the transition.  However, it'll still be a new workflow to learn, and that always requires some effort. 

What you can do right now to help is to put the main project page on your watchlist at Meta-Wiki and to make sure that the owners of your favorite tools and scripts are aware of this change and ready to make any necessary updates in December or January.  I don't expect this change to be visible here until April 2024 (and probably later than that), but this is a major change, and you shouldn't wait until the last minute.  Tell your friends, watch for information, and be ready to adjust.  If we all work together, we'll get through this, and at the end of that path, we should have better security for us plus better privacy for some contributors.

And, finally:  Thank you.  Thank you for answering my questions, thank you for making suggestions about software changes, and thank you for helping make the English Wikipedia truly one of the last best places on the internet.  Volunteer-me will doubtless be back next month with more questions. ;-) Whatamidoing (WMF) (talk) 20:27, 25 September 2023 (UTC)Reply[reply]

We look forward to sucking you in to some of our volunteer projects. I'm tempted to say something snarky as I often do about unfinished projects, but I recognize that if any of us were time-traveled back ten or fifteen years to the Wikipedia that was in production then, we sure would miss some of the progress that has been made. Thanks for working to communicate with the mess that is the tribe of technical editors here at en.WP. – Jonesey95 (talk) 21:43, 25 September 2023 (UTC)Reply[reply]
Many kudos for all the work you've done, @Whatamidoing (WMF), and I echo Jonesey's sentiment! {{u|Sdkb}}talk 16:40, 26 September 2023 (UTC)Reply[reply]

Add a "year" parameter with a subcategorization function to Template:Non-free biog-pic Edit

I propose to add a "year" parameter with a subcategorization function to Template:Non-free biog-pic, so that rather than classifying 25,000+ images in a single Category:Non-free biographical images, images would be categorized into, e.g., Category:Non-free biographical images published in 1939. This would be useful because all of these images will eventually fall into the public domain on a specific date. Publication year is a required parameter for uploading an image, so both the recategorization and the creation of the subcategories could be largely automated. It would only take someone with the template-editing skills to implement. BD2412 T 01:53, 26 September 2023 (UTC)Reply[reply]

Please keep discussions in one place. A version of the above was posted at Template talk:Non-free biog-pic. – Jonesey95 (talk) 01:58, 26 September 2023 (UTC)Reply[reply]
Yes, I posted that, and immediately saw in the page history that no one has been responding to discussions posted there for years. BD2412 T 02:14, 26 September 2023 (UTC)Reply[reply]

Why are there big, red Lua "not enough memory" messages throughout Comparison of web browsers? Largoplazo (talk) 09:56, 26 September 2023 (UTC)Reply[reply]

Because the server ran out of memory budget to process the page. Reduce the size/complexity of the page. —TheDJ (talkcontribs) 09:59, 26 September 2023 (UTC)Reply[reply]
Is this a problem with {{wikidata}}? According to the NewPP limit report (accessible near the bottom of the rendered page's HTML source) calls of that template take comparatively most of the time to build the page (not sure about memory use). —Kusma (talk) 10:17, 26 September 2023 (UTC)Reply[reply]
The problem might actually be the length of some of the Wikidata pages being used, compare Module_talk:Wd/Archive_1#COVID-19_pandemic_in_Luxembourg_totally_broken. —Kusma (talk) 10:26, 26 September 2023 (UTC)Reply[reply]
Taking the single row for the Edge browser as an example, it has ten calls to {{wikidata}} and that one row uses 12MB of Lua memory out of the total allowed 50MB. Not going to work. Some investigation of the history article would be interesting—do old versions have this problem? Has anything else recently changed? Johnuniq (talk) 10:51, 26 September 2023 (UTC)Reply[reply]
The Wikidata item for Edge, d:Q18698690, has grown from ca. 180k to over 500k in the last three years. Seems we can't use this item anymore. It is a bit ironic that Wikidata's accumulation of more and more data is making that data unusable for Wikipedia. —Kusma (talk) 11:05, 26 September 2023 (UTC)Reply[reply]
"Comparison of web browsers" should use the |sep parameter in the wikidata template, in order to get one transclusion per row, instead of two. Since memory from one transclusion is not accessable to the second transclusion, it will help quite a bit. Snævar (talk) 16:28, 26 September 2023 (UTC)Reply[reply]
Could you try to add some information about memory use and how to make the calls more efficient to the template documentation? It is not obvious at all to the casual observer with little knowledge of Lua memory issues that this would make a difference. —Kusma (talk) 16:37, 26 September 2023 (UTC)Reply[reply]
That solution is probably not going to solve the issue as is, but it will help. When there are optimizations, often they both require using the functions less and using said optimization. The proposal is doing something like |sep=</td><td> and merge two transclusions that way. <td> is used and not || because lua does not process wikicode, unless asked specifically to do so. |sep= is in the documentation. I think mathing it out would be more effort than actually doing it, so I do not see the point. I am thinking about this logically, not mathematically. Snævar (talk) 16:39, 27 September 2023 (UTC)Reply[reply]

Tech News: 2023-39 Edit

MediaWiki message delivery 16:49, 26 September 2023 (UTC)Reply[reply]

Recently I Google searched some material that was added at EFFPR and found out that Google displays a result for that page - but now that I've thought about it and examined WP:EFFP I've found that EFFP has __NOINDEX__.

The rendered meta tag for both are:
- <meta name="robots" content="max-image-preview:standard"> for Wikipedia:Edit filter/False positives/Reports; and
- <meta name="robots" content="noindex,follow,max-image-preview:standard"> for Wikipedia:Edit filter/False positives.

My question is then:
- Is it actually intended for EFFPR to be displayed by search engines, or should it be like EFFP?

2804:F14:80FB:2E01:4CA1:2349:2447:241C (talk) 11:09, 27 September 2023 (UTC)Reply[reply]

Brackets in title in italics Edit

I have encountered an unusual technical problem and don't know what to do. The page name of the article about the Picasso painting Bust of a Seated Woman (Jacqueline Roque) should be all in italics because the name in brackets is an integral part of the title. But the page name actually displays the italics only outside of the brackets, which makes it appear as if Jacqueline Roque was the author of a painting called Bust of a Seated Woman. @Johnbod: have you had this problem before? Edelseider (talk) 15:56, 27 September 2023 (UTC)Reply[reply]

The instructions at {{Infobox artwork}} were a bit confusing. I have improved them. The infobox needed |italic title=all to force italics on the parenthetical portion of the name. – Jonesey95 (talk) 16:28, 27 September 2023 (UTC)Reply[reply]
Thank you so much! Edelseider (talk) 16:29, 27 September 2023 (UTC)Reply[reply]

Odd infobox issue Edit

I find myself unable to publish edits at John von Neumann which remove the lengthy "known for" list in the infobox; when I click the "publish changes" button after editing, the page just reloads with no change. ~~ AirshipJungleman29 (talk) 19:06, 27 September 2023 (UTC)Reply[reply]

Worked for me. —Cryptic 19:19, 27 September 2023 (UTC)Reply[reply]
What page reloads? Are you seeing the edit window again, or the saved, rendered page without your intended changes? If the former, have you enabled the preference to require an edit summary? I have that preference enabled, and when I try to save without an edit summary, I get the edit window again, with a tiny message that is easy to miss, reminding me to add an edit summary. – Jonesey95 (talk) 19:28, 27 September 2023 (UTC)Reply[reply]
The preference to remind for an edit summary is a godsend, the tiny little box is very annoying. -- LCU ActivelyDisinterested transmissions °co-ords° 20:17, 27 September 2023 (UTC)Reply[reply]