Jump to content

Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by AKlapper (WMF) (talk | contribs) at 16:59, 12 October 2020 (→Error message should not appear?). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

 Policy Technical Proposals Idea lab WMF Miscellaneous 
Shortcuts
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).

Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. 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.


Accessing archived references

I know that this isn't, strictly speaking, a WP issue, but it arises for me here because this (WP) is the only place i try and access archive.org pages. If i click on this link [https://web.archive.org/web/20180717013125/https://www.indomiliter.com/panhard-vbl-penjaga-rumah-boediono/] in List of equipment of the Indonesian Army i get a Unable to connect Firefox can’t establish a connection to the server at web.archive.org message from my browser. This happens with every single archive link i try and verify (in every article i try, not just the one i mentioned), yet i know the links are active because i've used IABot to find them.. I will be ever so grateful if someone can tell me why this happens and what i need to do to stop it, because there are a lot of archived references i'd like to work with. I expect it's something very simple, silly even, so don't be worried about offending me by pointing out my complete lack of technical skill or understanding. Thanks in advance; happy days, LindsayHello 18:07, 1 October 2020 (UTC)[reply]

@LindsayH:: are you on a filtered internet connection? It's relatively common for strict DNS-based web filters to block web.archive.org, because unless your computer is configured to cooperate with the filter, it's impossible to tell which archived page you're looking at. Vahurzpu (talk) 18:17, 1 October 2020 (UTC)[reply]
Can you get to https://web.archive.org/ at all? If not, try it on a different device (like a phone or a computer), on a different internet connection (e.g. if you are having trouble on wireless, try a phone without wireless). Also try a different web browser (Chrome, Safari). – Jonesey95 (talk) 18:21, 1 October 2020 (UTC)[reply]
Anecdotally, I had that problem for a little while with one browser (though links worked in other browsers), then after awhile the problem went away. I surmised it was a browser problem that had gotten fixed on the back end. Schazjmd (talk) 18:24, 1 October 2020 (UTC)[reply]
Vahurzpu, i don't even know what a filtered internet connection is. My laptop is plugged in directly to my router which is plugged into my broadband and i don't think i've got any software limiting what i can do or where i can go.
Jonesey95, i cannot connect to https://web.archive.org/ at all, neither in Firefox nor in Chrome (Chrome says This site can’t be reached web.archive.org refused to connect, which i assume is essentially the same as the Firefox message). I can get to the site on my phone, which uses a wireless connexion to the same router; i guess i can use the phone to look up what i need and then type the entries into the browser on the laptop; less than ideal, but it should work. Thanks for the help and suggestions; happy days, LindsayHello 19:57, 1 October 2020 (UTC)[reply]
What operating system? --Izno (talk) 20:32, 1 October 2020 (UTC)[reply]
Windows 10, presumably the most recent version since MS does that cool "We're gonna update you whether you're ready or not" thing; Firefox 80.0.1. And the phone, which does work, is Android 9, which claims it is the current and up to date; happy days, LindsayHello 21:30, 1 October 2020 (UTC)[reply]
Lindsay, is it all of archive.org or just the web archives (I see you can't get to the subdomain above, but there are other parts of the site)? Do you live in a country with any marginally censorious regime? Have you tried uninstalling and reinstalling one of the browsers of interest (I'm wondering now if it is possibly one of your security certificates became corrupted; I don't know why that would impact just Archive.org). When your mobile can access it, is it connected wirelessly to your home router? I think the only other possibility pending/confirmed by those answers is that your ISP is blocking access. --Izno (talk) 02:47, 4 October 2020 (UTC)[reply]
Izno, it is not the whole of archive.org; i just went to https://archive.org/details/brooklynmuseum-o4368-the-great-red-dragon-and-the-woman for example, and it worked perfectly. I have uninstalled and then reinstalled Chrome, and still have exactly the same issue, the "refused to connect" message. Strangely, my phone won't now connect, so i don't know if it did previously or if i did something wrong ~ clicked an incorrect URL and thought it was on the archive site but not. I live in a country with no restrictions that i know of. My ISP is a very large, popular one ~ guaranteed i'm not the only person editing using it, probably not the only person on this page at the moment. You mention something about security certificates; is there something i can check or change there? I do appreciate the help; happy days, LindsayHello 06:58, 6 October 2020 (UTC)[reply]
We're past what I know on that point. One other thing I thought of was that you have some sort of ad or script blocking, but I'm pretty sure it's not that given you can access the other page. I am not sure who would be better to help with this. --Izno (talk) 16:23, 8 October 2020 (UTC)[reply]

Technical volunteer needed for Arbitration Committee Elections

Due to an RfC update, the candidates page which typically randomizes the transclusion order of ArbCom candidates, should only randomize once for a given user and then stay fixed. The easiest way to implement this would be some form of hashing and then sorting by the hash, which will attain psuedo-randomness per user and then stay fixed for each user. This requires a Lua module, and I am not very experienced with Lua. If anyone is up to the task, please let me know, so I can work with you on getting this set up for this year's elections.—CYBERPOWER (Around) 14:08, 3 October 2020 (UTC)[reply]

I believe you are referring to the code at Wikipedia:Arbitration Committee Elections December 2019/Candidates#Standing candidates which appears as follows.
{{#invoke:random|list|separator=newline
|{{Wikipedia:Arbitration Committee Elections December 2019/Candidates/First candidate/Statement}}
|{{Wikipedia:Arbitration Committee Elections December 2019/Candidates/Second candidate/Statement}}
|{{Wikipedia:Arbitration Committee Elections December 2019/Candidates/Third candidate/Statement}}
|{{Wikipedia:Arbitration Committee Elections December 2019/Candidates/Fourth candidate/Statement}}
}}
That uses Module:Random which will render the list of statements in a different order each time the page is purged. There is no way Lua can make the list stay in a fixed order for each logged-in editor reading the page. I'm pretty sure Lua cannot discover the reader's user name, and even if it could, Lua has no mechanism to store per-user or any other kind of data. Some JavaScript might be able to handle that by putting it in the browser's local storage or a cookie, but I don't know if there is a precedent for a site-wide script to do something like that. The caching mechanism of MediaWiki is opposed to any procedure that attempts to show different text to different users. Johnuniq (talk) 23:21, 3 October 2020 (UTC)[reply]
@Johnuniq: Actually, there's one clever way it could be done. Much of the MediaWiki interface is re-rendered for each user, and within it, {{REVISIONUSER}} is the user who's looking at the page. We could seed the RNG with the username. The only tricky part is figuring out whether or not we have somewhere suitable in the interface where we can inject such code without affecting the rest of the site. Jackmcbarn (talk) 23:27, 3 October 2020 (UTC)[reply]
Thanks, I'd forgotten about the magic words. There is some confusion because Help:Magic words#Other variables by type has a note "This shows the last user to edit the page. There is no way to show the user viewing the page due to technical restrictions." Here is a test:
  • User = {{REVISIONUSER}} → User = AKlapper (WMF)
Even if that worked, the next step would be tricky. Johnuniq (talk) 23:42, 3 October 2020 (UTC)[reply]
(edit conflict) Doesn't {{REVISIONUSER}} store the user who last edited the page, not the user who is currently viewing it? * Pppery * it has begun... 23:43, 3 October 2020 (UTC)[reply]
@Pppery: On regular pages, it's the user who last edited the page. But in parts of the interface, it is the user who's currently viewing it. Jackmcbarn (talk) 02:50, 4 October 2020 (UTC)[reply]
Correction: I'm almost positive this worked somewhere at some point, but now when I try it on testwiki, it just gives me the empty string. Back to the drawing board I guess. Jackmcbarn (talk) 02:56, 4 October 2020 (UTC)[reply]
For your first comment: I assume by "regular pages" you mean editable wiki pages, for that what you said is true, but the "interface" part is not correct. MediaWiki: namespace pages are considered part of the interface and they are just editable wiki pages. So the magic word will behave the same on MediaWiki:Example as it will do here. {{REVISIONUSER}} is based on revision. On virtual pages (such as SpecialPages where there are no revisions) it's null actually, and outputs empty string.

The only place where it's always the user doing the edit, is during 'edit previewing', that's before saving the actual edit, and that area is exclusive to the concerned user. Once it's saved, it's then either dynamic (shows the last revision user always), or it'll become hardcoded string username of the user who did the actual edit with it because of subst modifier. As far as I know, there's no magic word that refers to the current viewing user always. JS world does have some kind of that, not exactly though. It's called "relevant user".

For the second comment: You are probably thinking of {{subst:REVISIONUSER}}. – Ammarpad (talk) 16:52, 4 October 2020 (UTC)[reply]
As I commented on the RFC, this requires either a Javascript or a PHP implementation (Lua does not and will likely never know who is viewing a page), and I already discussed that probably neither option is a good other option. It is unfortunate my comment was not paid heed to. --Izno (talk) 02:12, 4 October 2020 (UTC)[reply]
Izno, I closed the RfC in favor of this implementation, because I believed the implementation of it was feasible. If it turns out it isn't, then I will have to amend the close. —CYBERPOWER (Chat) 22:38, 5 October 2020 (UTC)[reply]
I see the point of your opening comment about hashing now—using a hash of the current user's name concatenated with the candidate's user name as a sort key would suffice for a random order and would not require the storage of any per-user information. However, as Izno says, only PHP or Javascript can get the name of the user reading the page. PHP is not going to happen but Javascript is feasible. Except, I think the Javascript would have to be loaded for all users on all pages and that is also not going to happen. Accordingly, I think the current code (random order on each page purge) will have to do. Johnuniq (talk) 22:56, 5 October 2020 (UTC)[reply]
Cyberpower678 Hah, I was not chastising you. I know you did due diligence as closer. :) --Izno (talk) 23:00, 5 October 2020 (UTC)[reply]
Izno, :-) —CYBERPOWER (Around) 23:06, 5 October 2020 (UTC)[reply]
I thought everyone knew that this would require JS to implement. We could install it temporarily as a default-enabled gadget, couldn't we? Best, Kevin (alt of L235 · t · c) 00:55, 6 October 2020 (UTC)[reply]
If so, the gadget code should be wrapped in a loader that is only activated on the correct page(s), to avoid always loading the code on all page views. I can take a crack at writing the js this week(end) if no one beats me to it DannyS712 (talk) 05:29, 6 October 2020 (UTC)[reply]
DannyS712, the sooner, the better. We only have 2 weeks to have a working solution implemented. —CYBERPOWER (Chat) 15:22, 6 October 2020 (UTC)[reply]
...I have real life, sorry DannyS712 (talk) 15:49, 6 October 2020 (UTC)[reply]
@Cyberpower678: there are no deadlines -- just because the RfC said they want this, doesn't mean that it will be willed in to existence - is the fall back if this isn't created to go back to the prior year's random scheme? — xaosflux Talk 18:03, 8 October 2020 (UTC)[reply]
Xaosflux, sure it is, but I would ideally like to have the RFC implemented as requested if possible. —CYBERPOWER (Chat) 18:50, 8 October 2020 (UTC)[reply]
Wikipedia:Interface administrators' noticeboard/Archive 1 § Chess viewer has some discussion on how to implement a loader gadget that would load a gadget on demand for specific pages. User:קיפודנחש (aka kipod) may be interested in helping out, as this would be useful to implement the consensus reached at Wikipedia:Village pump (technical)/Archive 175 § Enable chess PGN viewer for chess articles. Also pinging User:Wugapodes. isaacl (talk) 16:31, 7 October 2020 (UTC)[reply]
will be delighted to assist. i do not have permissions, so currently can't help with actual implementation, but i will give any support i can, time permitting. the mechanism was originated at ruwiki (i believe). we adopted it in hewiki, and added support for gadgets. we wanted gadgets because RL is more efficient with them, e.g. by using minimized code. the mechanism is made of one template, say, Template:Load script, and a dozen or so lines in common.js. for clarity, let me copy the js code here:
// On demand loading of scripts and gadgets, initial version from ruwiki.
// Detects uses of template "טען סקריפט"
// and loads specifically-named scripts or gadgets.
// for a script to be loadable this way, its name must begin with "Mediawiki:Scripts/"
// for a gadget, its name as defined in gadgets-definition must begin with "ondemand-"
if ( mw.config.get('wgCanonicalNamespace') !== 'Special' ) 
mw.hook( 'wikipage.content' ).add( function( content ) {
		var beenthere = {};
		$( '.executeJS', content ).each( function () {
			var script = $( this ).data( 'scriptname' );
			if ( script && $.trim( script ) ) {
				script = $.trim( script );
				if ( ! beenthere[script] )
					mw.loader.load( "/enwiki/w/index.php?title=Mediawiki:Scripts/" + script + ".js&action=raw&ctype=text/javascript", "text/javascript" );
				beenthere[script] = true;
			}
			var gadget = $( this ).data( 'gadgetname' );
			if ( gadget && $.trim( gadget ) ) mw.loader.load( 'ext.gadget.ondemand-' + $.trim( gadget ) ); // np repetitions - resourceloader takes care
		} );
} )
so we scan the page for a special signature, ".executeJS", which is left by the template. the template "encodes" the name of the desired script, and the code above extracts it, and tries to load the script or gadget. we believe this is safe from exploitation - for instance, the can not be used to load just "any" gadget - only gadgets whose name begins with "ondemand-". those gadgets are hidden, and were deemed "safe" by the interface editors. similarly, only scripts whose name begins with "mediawiki:Scripts/" can be run this way, and again, this "namespace" is reserved for scripts approved by interface admins. let me know if there's anything i can help with. peace - קיפודנחש (aka kipod) (talk) 19:59, 7 October 2020 (UTC)[reply]
Thanks for the code and explanation. Dreamy Jazz talk to me | my contributions 20:28, 7 October 2020 (UTC)[reply]
@קיפודנחש: I found a way to exploit it: if the script name is &title=User:Mr._Stradivarius/evil.js&whatever=, then the URL passed to mw.loader.load will be /enwiki/w/index.php?title=Mediawiki:Scripts/&title=User:Mr._Stradivarius/evil.js&whatever=.js&action=raw&ctype=text/javascript, meaning User:Mr. Stradivarius/evil.js will be loaded. (That page is a red link now, but can be created with a suitably evil script if desired. ;) — Mr. Stradivarius ♪ talk ♪ 12:19, 9 October 2020 (UTC)[reply]
@Mr. Stradivarius: - thanks. looking at it now. will ask our local expect to fortify it to prevent this specific explit. peace - קיפודנחש (aka kipod) (talk) 16:22, 9 October 2020 (UTC)[reply]
@קיפודנחש and ערן: It should be OK if you encode the script name using encodeURIComponent. I would encode the gadget name as well, just to be sure. That will help to protect against possible future changes to mw.loader.load itself. — Mr. Stradivarius ♪ talk ♪ 22:36, 9 October 2020 (UTC)[reply]
This use case is a far stretch from requiring adding javascript for every page view -- maybe a Gadget for rights=purge (to only target logged in users), but that is still a lot of overhead to force on every single page load, I brought up in the RfC that the desire to do something was being put in front of the work required to actually do it - so while the RfC can still say this is desirable, so are 10 years of backloged phab tickets - someone will want to really work on this. As far as page-specific scripts, yes I think that is a good idea, but shouldnt require running scripts on every page. Oh look, here is phab:T8883 from 2006 if someone really wants to work on per-page scripting! — xaosflux Talk 18:01, 8 October 2020 (UTC)[reply]
@Cyberpower678 do you have a mockup or example that I can use with to develop the script (i.e. a page transcluding multiple subpages that have different content, so I can test showing them in a random order)? DannyS712 (talk) 02:21, 9 October 2020 (UTC)[reply]
@DannyS712: I created a sandbox from last years: Wikipedia:Arbitration Committee Elections December 2019/Candidates/Sandbox, for the most part we'd expect the page to stay the same other than the {{#invoke:random|list|separator=newline section. — xaosflux Talk 12:58, 9 October 2020 (UTC)[reply]

How to do three brackets?

If there need to be single brackets and the first bracket is followed immediately by a Wikilink, is there a way to keep from having a space between them?— Vchimpanzee • talk • contributions • 18:12, 4 October 2020 (UTC)[reply]

[<nowiki/>[[foo]]] ([foo]) works. * Pppery * it has begun... 18:13, 4 October 2020 (UTC)[reply]
{{!(}} is documented to produce the open square bracket (just like {{Square bracket open}}) but it just uses the actual character so {{!(}}[[foo]] does not do what you want. That seems like a bug. Maybe it should use &#91; ?DMacks (talk) 16:13, 6 October 2020 (UTC)[reply]
Those templates are often used for creating links, tables, etc. where direct input doesn't allow it, so replacing them with numeric character references will probably break lots of templates. Nardog (talk) 08:31, 7 October 2020 (UTC)[reply]
A help desk question I just saw was about the same thing.— Vchimpanzee • talk • contributions • 17:58, 8 October 2020 (UTC)[reply]

How to show a live number of English Wikipedia articles on another Wikimedia project?

Hi all

I'm writing some instructions on Wikidata and would like to add a live number of articles on English Wikipedia to them. I know about {{NUMBEROFARTICLES}} which gives 6,938,981 but is it possible to get something similar to {{NUMBEROFARTICLES}} on another Wikimedia projects? Not sure if you can use templates across projects?

Thanks

John Cummings (talk) 14:32, 5 October 2020 (UTC)[reply]

No, you cannot access that information on a different wiki. --Izno (talk) 14:45, 5 October 2020 (UTC)[reply]
That's true, but {{NUMBEROF}} now shows data that is updated four times a day. The following shows values which I obtained just before posting this:
  • {{NUMBEROF|articles|fr|N}} → 2,254,469
  • {{NUMBEROFARTICLES}} → 2 254 493 (at frwiki)
GreenC has a bot which does the updating, and has updated many projects with the new NUMBEROF code. Johnuniq (talk) 23:05, 5 October 2020 (UTC)[reply]
Thanks very much @Johnuniq:, would it be possible for this to work on Wikidata to show the number of English Wikipedia articles? Currently the template doesn't exist. John Cummings (talk) 08:35, 8 October 2020 (UTC)[reply]
Yes, but I can't do it now. I'll ping GreenC again to see if they are available. Otherwise I'll look on the weekend or maybe sooner. Remind me if I forget. Johnuniq (talk) 09:08, 8 October 2020 (UTC)[reply]
Thanks you very much for all your help @Johnuniq:, really appreciate it. John Cummings (talk) 09:44, 8 October 2020 (UTC)[reply]
@John Cummings: now installed: https://www.wikidata.org/wiki/Template:NUMBEROF -- GreenC 18:16, 8 October 2020 (UTC)[reply]
@GreenC: thank you so much, really helpful, if there's anything non technical I could help with let me know. John Cummings (talk) 21:08, 8 October 2020 (UTC)[reply]

Adapting nested lists to mobile

Please take a look at a page with nested lists like this on your mobile phone.

Sub-bullet-points currently cause the text to move to the right so that it's barely readable on small screens, with only very little space for the text that is squished to the right.

Which solution do you suggest or prefer?

Nested lists probably have to adapt to the mobile display by getting displayed differently in the mobile view (en.m.wikipedia) at least if it detected a small screen or a mobile device in general. One possible solution would be simply displaying the nested header above the content (either view these sub-bullet-points or treestyle view that could look similar to or even use Template:Category tree). The mobile view is important and should not be neglected.

I created this entry as you may be interested in the main discussion here.

--Prototyperspective (talk) 10:12, 7 October 2020 (UTC)[reply]

edit requests at Talk:2020 Nagorno-Karabakh conflict

Could somebody take a look at this page? My browser outright refuses to load this talk page, I suspect due to either some form of code malformation or sheer length. Unsurprisingly due to the current armed conflict, there has been a flurry of edit requests and I can't try to answer any at this time. Thanks in advance to anyone that can help. Eggishorn (talk) (contrib) 18:30, 7 October 2020 (UTC)[reply]

@Eggishorn: the page is quite large, try loading it in mobile view. — xaosflux Talk 18:41, 7 October 2020 (UTC)[reply]
@Xaosflux:, thanks for the tip, I always forget that I use the mobile website address easily while on desktop. Unfortunately it means that I can't use tools like one-click archiver or edit request helper, though. I see from looking at the edit requests that long-time editors there seem to be adequately engaging with the requesters so I will just let them handle those. Thanks again. Eggishorn (talk) (contrib) 18:48, 7 October 2020 (UTC)[reply]
@Eggishorn: another option you could try is safemode - however this will also disable all gadgets and scripts, but it helps if one of those is the reason the page won't load. — xaosflux Talk 18:56, 7 October 2020 (UTC)[reply]

WP:PCT

For the past few days, the page curation toolbar has strangely not been loading for me. Can anyone help me? I'm finding NPP difficult without the toolbar. JavaHurricane 06:00, 8 October 2020 (UTC)[reply]

Cannot verify my email address

I have had my account for years. Several weeks ago, I received a Wikipedia alert which said, "Your registered email address, xxxx@xxxxx,com, has been unsubscribed due to multiple message delivery failures. You can verify your email address again." Two days later, I stopped receiving Wikipedia emails. I have attempted on several occasions to verify my email address, but I never receive any email back from Wikipedia. What can I do to remedy this?--Quisqualis (talk) 08:15, 8 October 2020 (UTC)[reply]

@Quisqualis: This may be related to this problem experienced by llywrch recently. —[AlanM1 (talk)]— 14:45, 8 October 2020 (UTC)[reply]
Thanks for the pointer, AlanM1. I found that by removing and then immediately replacing my email address, the problem of not being verified has been solved.--Quisqualis (talk) 22:44, 8 October 2020 (UTC)[reply]

Formatting ISO dates as per site/user preferences in JavaScript

The MediaWiki API returns timestamps in the ISO format (yyyy-mm-ddThh:mm:ssZ). Is there a module or something you can use in JavaScript that converts such a date according to whatever the site settings and user preferences on date format/timezone are, i.e. like the dates you see in page history, contributions, etc.? Nardog (talk) 09:08, 8 October 2020 (UTC) [reply]

Extended content
Try the #dateformat magic word — GhostInTheMachine talk to me 10:09, 8 October 2020 (UTC)[reply]
@GhostInTheMachine: Thanks, but that's far from what I'm looking for. It can't handle times (H:M:S), let alone timezones, does nothing if the user has no preference set (rather than use the site's default), has poor internationalization support ({{#dateformat:2020-10-08|ymd}} on zhwp returns "2020 10月 8", not "2020年10月8日 (四)", and {{#dateformat:2020-10-08|dmy}} on thwp "8 ตุลาคม 2020", not "8 ตุลาคม 2563"), and, above all, you'd have to make another API query to get the parsed result. Nardog (talk) 10:50, 8 October 2020 (UTC)[reply]
I generally use #time but that has a fixed format. The template {{time}} has more options — 14:19, 12 1月 2025 UTC / 14:19, 12 มกราคม 2025 UTC but it does not "officially" take a timestamp as input, although this sort-of works: 07:05, 2 11月 2018 UTC / 07:05, 2 พฤศจิกายน 2018 UTC. Sorry, not sure what else to suggest. Maybe post something at Template talk:Time? — GhostInTheMachine talk to me 11:32, 8 October 2020 (UTC)[reply]
Again, I'm looking for something ready-made I can use to convert ISO timestamps in accordance with the site locale/settings and user preferences in JavaScript (preferably without sending the timestamps to a server). I know I can probably use Moment or something and come up with an elaborate way to replicate what the MediaWiki interface does, I'm just asking if I'd be reinventing the wheel if I did so. Nardog (talk) 11:50, 8 October 2020 (UTC)[reply]
OK, for JS take a look at the CommentsInLocalTime gadget — it responds to local browser settings. You may be able to clone some of the code for your use. There is also a fork at User:Mxn/CommentsInLocalTime — GhostInTheMachine talk to me 12:30, 8 October 2020 (UTC)[reply]
For the nth time, I'm looking for a way to replicate the MW interface. That's not what that script does. Nardog (talk) 12:36, 8 October 2020 (UTC)[reply]
I don't know javascript, but this discussion seems relevant. More search results. – Jonesey95 (talk) 15:02, 8 October 2020 (UTC)[reply]
I think mw:ResourceLoader/Core modules#mw.user.options is the place, it just has basically no documentation as to the names of the options, much less which are available from Javascript. Krinkle Is that the place? --Izno (talk) 15:57, 8 October 2020 (UTC)[reply]
@Izno: mw.user.options.get('date') and mw.user.options.get('timecorrection') will give you the user preferences (like "dmy" or "ZoneInfo|-240|America/New_York"), but that's not enough information for a script to convert a date in the ISO format to "16:20, 8 October 2020" or what have you. Nardog (talk) 16:20, 8 October 2020 (UTC)[reply]
I'm not sure I understand what the issue is. Are you asserting that mw.user.options says "here are the options" and not "here is this user's selection"? I am otherwise fairly certain that you can implement what you're looking to since ISO format encodes the applicable time zone (which is Z, in yyyy-mm-ddThh:mm:ssZ, for Zulu time aka UTC). --Izno (talk) 19:32, 8 October 2020 (UTC)[reply]
Or is the part you're stuck on turning ISO into English? Date might help for that. Otherwise no, I do not think there is anything in the MediaWiki ecosystem that does that. --Izno (talk) 19:39, 8 October 2020 (UTC)[reply]
Again, what I'm looking for is something ready-made that automatically figures out the site settings and user preferences and turns an ISO timestamp (or Date object or Unix time) to the exact format you see in page history, contributions, etc., nothing else. Nardog (talk) 23:37, 8 October 2020 (UTC)[reply]
Otherwise no, I do not think there is anything in the MediaWiki ecosystem that does that. --Izno (talk) 00:29, 9 October 2020 (UTC)[reply]
I don't think the API even provides a way to identify the site-default date format used on the wiki. Also note that the timecorrection value you get from mw.user.options doesn't account for DST. In JavaScript gadgets, presenting such things like dates and numbers in an i18n/l10n-compatible way is just one big headache. – SD0001 (talk) 19:21, 9 October 2020 (UTC)[reply]
Thanks. It's a shame there's no access point to the date formatting for JS even though the formatted dates appear in so many parts of the interface. Nardog (talk) 12:20, 10 October 2020 (UTC)[reply]

Odd one: category exists or not?

Take a look at Category:Wikipedia sockpuppets of Viraj Alom. Each user's userpage is identical, unless I'm missing something, which I probably am, because User:Tamilreporter123 insists on telling me 

"Category:Wikipedia sockpuppets of Viraj Alom does not exist. Please Use this link to create the category page. (The page will be pre-loaded. All you need to do is save it)"

Puzzled. --jpgordon𝄢𝄆 𝄐𝄇 19:00, 8 October 2020 (UTC)[reply]

 Works for me @Jpgordon: are you still seeing ths? Could you reload the page / purge your cache? — xaosflux Talk 19:11, 8 October 2020 (UTC)[reply]
I'm not seeing any problem either. But, for future reference, there's also caching that's happening on the server side, and the symptom you describe is a classic example of incorrect server-side cache invalidation. You can fix that by going to Special:Purge. In this specific case, https://en.wikipedia.org/enwiki/w/index.php?title=User:Tamilreporter123&action=purge would be what you want. -- RoySmith (talk) 19:18, 8 October 2020 (UTC)[reply]
Yeah, caching. I cleaned the cache on my side to no avail, but it's resolved itself since. --jpgordon𝄢𝄆 𝄐𝄇 22:55, 8 October 2020 (UTC)[reply]

Testing article issue detection AI

Help us test our problematic statement detection system. We hope to deploy this to help editors.

We are developing an AI to automatically detect issues in articles related to: NPOV, CLARIFY and CITE. We need help evaluating how well the model is working. We are asking for a group of volunteers to evaluate a set of sentences that are flagged by the AI. The landing page for evaluations can be found here. This page has a small set of examples for each issue. We have included sub-pages that include more examples (e.g. More POV examples). If you want to help, please assess as many example statements as you can. The more assessments we get, the better we can judge our model and make improvements. A description of our research project is on meta. Sumit (talk) 22:43, 8 October 2020 (UTC)[reply]

I'm working with Sumit on this project. You might remember some of the related work we did with mw:ORES vandalism detection and topic routing. We think this will also be a useful tool. Thanks in advance for your help! --EpochFail (talk • contribs) 22:48, 8 October 2020 (UTC)[reply]

Dark theme

Hello Wikipedians :)

Wouldn't it be great to have a dark theme? I quickly tinkered something here as a little foretaste. Could someone tell me if something like this is planned or who is responsible for it? To make a beautiful design, more beautiful than the current "Vector" is really not difficult! If there are people here who like the dark theme more than the current one, they could write code together, or not?

PS: if someone is seriously interested, you can write me an email, as I am not very active here. -Killarnee (C•T•U) 23:03, 8 October 2020 (UTC)[reply]

@Killarnee: you may want to follow or contribute at phab:T199634 as well to help build a dark vector theme for everyone. — xaosflux Talk 23:23, 8 October 2020 (UTC)[reply]

Searching for cases of multiple templates on one page

I there a tool, or some search wizardry, that I can use to find all articles that have both of two given templates on them?

For example, I know that Ba-ta-clan features both {{Offenbach operas}} and {{Jacques Offenbach}}. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:34, 9 October 2020 (UTC)[reply]

You can try with petscan. --Gonnym (talk) 14:43, 9 October 2020 (UTC)[reply]
@Pigsonthewing: this is the list of articles that transclude both of those templates:
List
Did that semi-manually by: Exporting the what-links-here transclusions, dumping them to a file, sorting, then using a regex to only keep duplicate lines: ^(.+)\r?\n(?!(.|\r?\n)*\1). — xaosflux Talk 14:45, 9 October 2020 (UTC)[reply]
Thank you, but that was a simple example; my use-case is for more highly-used templates. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:10, 9 October 2020 (UTC)[reply]
Special:Search. You may also use a minus symbol to identify pages without one or the other. --Izno (talk) 14:54, 9 October 2020 (UTC)[reply]
That's what I wanted. Thank you. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:10, 9 October 2020 (UTC)[reply]

I've hit a snag, in that I am getting results which include things like {{One source|section}} (see this example, which includes Pretoria#Jacaranda city). Is it possible to exclude such section-specific cases? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:39, 9 October 2020 (UTC)[reply]

@Pigsonthewing: using search, you can try a not-in-source argument as well, such as this. — xaosflux Talk 16:44, 9 October 2020 (UTC)[reply]
It should be hastemplate:"More citations needed" hastemplate:"One source" -insource:/\{\{One source\|section/ (the pipe in the regex has to be escaped). – SD0001 (talk) 18:38, 9 October 2020 (UTC)[reply]
You must consider redirects and whitespace also. --Izno (talk) 19:30, 9 October 2020 (UTC)[reply]

Grant request for localising Twinkle to other wikis

I have requested a m:Rapid Grant to add internationalisation and localisation support to WP:Twinkle (which would make it easy to port the gadget to other wikis). Any comments and community endorsements would be welcome at the Grant request page. This task has been stalled for over 8 years. Now may be the time it actually happens! See GitHub issue for the detailed plans on the work to be done. Regards, – SD0001 (talk) 18:41, 9 October 2020 (UTC)[reply]

Block log has talk page link when more than one IP is blocked

I was curious why someone with no contributions other than a Help Desk question would be blocked and I found this. There shouldn't be a talk page link for an IP that doesn't exist.— Vchimpanzee • talk • contributions • 22:01, 9 October 2020 (UTC)[reply]

Vchimpanzee, per the talk page history, it was created in error. I have tagged it for speedy deletion. Home Lander (talk) 22:15, 9 October 2020 (UTC)[reply]
Is there not some way to keep a link to the talk page from appearing?— Vchimpanzee • talk • contributions • 15:25, 10 October 2020 (UTC)[reply]

code error: out of place lines of text in the image of "Table of elements"

Resolved

Description is there: Talk:Chemical_element#code_error:_out_of_place_lines_in_the_image_of_"Table_of_elements"
Steue (talk) 22:27, 9 October 2020 (UTC)[reply]

This edit produced the page that begins "Your edit includes new external links. These may be much welcomed links to references" and requires a CAPTCHA to be passed. That wasn't a problem for me, but I've seen it happen in the past, and I now note that that page does not link to any way to report false positives, comparable to WP:Edit_filter/False_positives/Reports. I suggest (1) that it should, and (2) and the error behind this false positive be corrected.

There is no reason to reply to me about this; please discuss it here if you want, or move this report to a better forum if you know of one, and either take the actions or not. Thanks for your attention. --174.89.48.182 (talk) 06:11, 10 October 2020 (UTC)[reply]

Possibly, there's a malformed link already on the page that the software thinks is "new" with every edit (as in phab:T223195). As no one seems interested in fixing that bug, probably best just to remove the link, though I don't want to hunt for it right now.
Perhaps the message should link to this page (WP:VPT)? Or would we end up with too much garbage from drive-by spammers? Suffusion of Yellow (talk) 06:23, 10 October 2020 (UTC)[reply]
The link to http://www.recordonline.com/apps/pbcs.dll/article is probably the problem -- it redirects about 7 times and then redirects in a circle — GhostInTheMachine talk to me 15:52, 10 October 2020 (UTC)[reply]
Actually, I couldn't reproduce this. So maybe a one-off glitch. That link shouldn't cause a problem; the servers don't visit any of the links before you save (though some bots might afterwards). Suffusion of Yellow (talk) 00:41, 11 October 2020 (UTC)[reply]

Extract a domain from a URL with Lua

I want to get the domain (like aberdeennews.com) from a URL (like http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member). mw:Extension:Scribunto/Lua reference manual#Patterns is nice enough to mention that character escaping is not supported (so I must make an ugly workaround for that), alternation is not supported (to make my life harder I presume), everything is greedy (to make my life a living hell I presume) and in the end, I just can't figure it out whatever I try.

The closest I got was {{#invoke:String|match|s={{#invoke:String|replace|source=http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member|pattern=.|replace=@|count=all|plain=true}}|pattern=([^/@]*)@([^@/]*)/|start=1|match=1|plain=false|nomatch=not_found}} (aberdeennews) and it doesn't work. (also it's very ugly) — Alexis Jazz (talk or ping me) 13:31, 10 October 2020 (UTC)[reply]

This can be done in Lua with mw.uri.new("http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member").host I can't find any Wikitext interface to that functionality, though. * Pppery * it has begun... 14:07, 10 October 2020 (UTC)[reply]
@Pppery: Thanks. I found a way now, but it's not very pretty. I'd prefer the function you provided, but I'm too close to snapping. I tried to create a small module with the code you gave me. I shouldn't have. I really shouldn't. I need to go do something else for the sake of my blood pressure. — Alexis Jazz (talk or ping me) 15:03, 10 October 2020 (UTC)[reply]
You can get the same result as mw.uri.new(url).host with string.match( url, "%/%/(.-)%/" ). In Lua % is the escape character and - is the only non-greedy quantifier.
So in a test function {{#invoke:Sandbox/Jts1882/Test|getdomain|http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member}} generates articles.aberdeennews.com|articles.aberdeennews.com (the result generated by the two methods separated by a pipe). —  Jts1882 | talk  16:25, 10 October 2020 (UTC)[reply]
As wikitext, {{#invoke:String|match|http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member|%/%/(.-)%/|plain=false}} -> articles.aberdeennews.com. * Pppery * it has begun... 20:30, 10 October 2020 (UTC)[reply]
@Jts1882 and Pppery: Thank you both! Good to learn about the %. However, I specifically need the domain (aberdeennews.com), not the subdomain. — Alexis Jazz (talk or ping me) 23:45, 10 October 2020 (UTC)[reply]
Does {{#invoke:String|match|articles.aberdeennews.com|[^.]-%.[^.]-$|plain=false}} -> aberdeennews.com work? * Pppery * it has begun... 00:15, 11 October 2020 (UTC)[reply]
You can try this to extract the last two components of the host name: {{#invoke:String|match|http://articles.aberdeennews.com/2007-09-23/news/26395715_1_kosher-meat-aberdeen-congregation-member|//.-([^./]+%.[^./]+)/|plain=false}} -> aberdeennews.com isaacl (talk) 00:25, 11 October 2020 (UTC)[reply]
@Pppery: That doesn't remove what comes after the TLD and doesn't work with http://aberdeennews.com/article.
@Isaacl: That appears to work, and looks much better! Thanks! — Alexis Jazz (talk or ping me) 00:38, 11 October 2020 (UTC)[reply]

If you are looking for .com domains, a string search will work. In general, however, it is not possible to extract the domain from a URL without a look-up table such as that described here. For example, the domain of aa.bb.com is bb.com but the domain of aa.co.uk is aa.co.uk. Consider the URL http://www.news.kku.ac.th/newspaper/content/view/545/86/ (from Christian Wolther). The host is www.news.kku.ac.th and the domain is kku.ac.th. Or, try http://www.portoacre.ac.gov.br/dados-municipio (from Porto Acre) which has domain portoacre.ac.gov.br. [I edited this comment to correct an error.] Johnuniq (talk) 03:58, 11 October 2020 (UTC)[reply]

Thanks for raising this aspect. I was going to comment but the page you linked to provides much more detail. isaacl (talk) 04:16, 11 October 2020 (UTC)[reply]

IP talk pages: how many?

How many user IP talk pages do we have? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:08, 10 October 2020 (UTC)[reply]

Could probably be answered by someone using Quarry, but I can't. — Alexis Jazz (talk or ping me) 00:28, 11 October 2020 (UTC)[reply]
@Pigsonthewing: 9,280,186 (quarry:query/48920). – SD0001 (talk) 13:00, 11 October 2020 (UTC)[reply]
Of course the count doesn't include the 500,000+ IP talk pages which MZMcBride deleted back in the day when such deletions were considered legit. – SD0001 (talk) 13:07, 11 October 2020 (UTC)[reply]

Very helpful, thank you. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:00, 11 October 2020 (UTC)[reply]

js edit needed to clear renamed category

Resolved
 – The page was updated, just as with other protected pages an edit request would also suffice for such requests in the future. — xaosflux Talk 15:43, 11 October 2020 (UTC)[reply]

Category:Wikipedians who use IRC was renamed to Category:Wikipedians who use Internet Relay Chat; however the old name still has two entries due to code on User:Yamakiri/Userpage2.js. The user hasn't edited Wikipedia in six years so can someone with access sort out the changeover please? Thanks in advance. Timrollpickering (talk) 21:40, 10 October 2020 (UTC)[reply]

@Timrollpickering: I've added a note about this discussion at the Interface administrators' noticeboard. Graham87 05:50, 11 October 2020 (UTC)[reply]

That page contains zero valid javascript and should be renamed to something without the suffix .js. Clever trick to make his userpage content uneditable by others, but it probably shouldn't be allowed (for exactly this reason). ―cobaltcigs 08:58, 11 October 2020 (UTC)[reply]

Editing thumb style images via templatestyles

I'm trying to edit the basic display of thumb images using templatestyles. However I've not seemed to get it working. Am I using the wrong syntax on my styles.css page? See my test here:

Thanks in advance for any advice. T.Shafee(Evo&Evo)talk 01:39, 11 October 2020 (UTC)[reply]

@Evolution and evolvability: A few notes:
  1. It's the thumbinner class that has a border, not the thumb class, so if you want to recolor it, you have to use that one.
  2. To use .ns-0 or other classes on <body>, you need to use "body.ns-0" to keep MediaWiki from sticking ".mw-parser-output" on the wrong side of it and breaking it. (Ignore the warning that it's overqualified, see phab:T254671.)
  3. Using * and #content like you are is unnecessary.
  4. Since white and #FFFFFF are the same color, your specialization for ns-0 isn't accomplishing anything useful. To see an effect, change one of them to a different color.
I fixed all but the last point for you. Jackmcbarn (talk) 19:19, 11 October 2020 (UTC)[reply]
@Jackmcbarn: Aha, thank you. Exactly what I needed to get me on the right track. T.Shafee(Evo&Evo)talk 11:14, 12 October 2020 (UTC)[reply]

Signature not working

I recently changed my signature so that it has my name, and links to my talk page and contribs page. Here is my signature: ―[[User:Sportzpikachu|''Sportzpikachu'']] <sup>[[User_talk:Sportzpikachu|<span style="color:green">talk to me</span>]]</sup><sub style="position:relative;left:-4.4em">[[Special:Contributions/Sportzpikachu|<span style="color:green">contribs</span>]]</sub> It all works properly, but there is a gap between my signature and the time, as you can see: ―Sportzpikachu talk to mecontribs 07:12, 11 October 2020 (UTC) Any tips on how to improve/fix my signature?[reply]

Try this (based partly upon {{sfrac}} template):
―Sportzpikachu talk to mecontribs 08:43, 11 October 2020 (UTC)[reply]
―cobaltcigs 08:43, 11 October 2020 (UTC)[reply]
At 338 characters that would violate WP:SIGLEN. --Redrose64 🌹 (talk) 10:08, 11 October 2020 (UTC)[reply]
How about this? "―Sportzpikachu talk to mecontribs" ―[[User:Sportzpikachu|''Sportzpikachu'']] <sub style=display:inline-block>[[User_talk:Sportzpikachu|<span style=color:green;display:block>talk to me</span>]][[Special:Contribs/Sportzpikachu|<span style=color:green;display:block>contribs</span>]]</sub> It weighs in at 253 characters. Jackmcbarn (talk) 19:05, 11 October 2020 (UTC)[reply]
@Jackmcbarn: Seems to work, although I had to replace talk to me with my talk to make room for the quotes. ―Sportzpikachu my talkcontribs 00:14, 12 October 2020 (UTC)[reply]

Non-superscript refs for tables

Commonly, we often use tables with a single column set aside for references for the given row, but using the standard <ref> format, these refs come up as superscript and seem to waste space. Do we have a means to create a non-subscript version of a reference (otherwise conforming to all the same properties as a normal ref including elements like name and group)? --Masem (t) 15:50, 11 October 2020 (UTC)[reply]

Directly in the ref tag? No. Templatestyles can do it by removing the superscript, but I would advise against it generally, and this use case is not compelling to suggest otherwise. --Izno (talk) 23:04, 11 October 2020 (UTC)[reply]
You could do it like this:
Table showing direct reference links
Header1 References
Statement Smith 2006

Sources

  • Smith, Jane (2006). Title.
If that is compatible with the table you are using. – Jonesey95 (talk) 23:53, 11 October 2020 (UTC)[reply]

Case-insensitive page names

I know that pages on Wikipedia are case sensitive, except for the first letter or the namespace prefix (which doesn't count as part of the page name.) For example, Wikipedia:Why create an account? and Wikipedia:WhY CreAte aN acCoUnT? are unequivalent. However, it seems that special pages break this rule. Ok, so "Special:" has a capital S and the first letter after the prefix needs to be capitalized. However, it seems that the entire page name of any special page is case-insensitive, for example: Special:CreateAccount and Special:CreAteacCoUNt are equivalent. Why is this? --Gioguch (talk) 22:41, 11 October 2020 (UTC)[reply]

Special pages are special :); moreover, they are not wikipages so the rules for wikipages do not apply. --Izno (talk) 23:07, 11 October 2020 (UTC)[reply]

Special:Nospecialpage

I don't know how much you know about code, but I am curious about what page has the message "No such special page." For example, I have the mindset tthat everything on English Wikipedia is a "page." A "page" is a page that is in any namespace, so https://www.wikipedia.org is not a page, it's just an HTML page unrelated to the MediaWiki software. The page Talk:Main Page is a page, and Talk: is really just the page Special:Badtitle. That's what page you get directed to when you log out. Any invalid special page name just redirects you to the name of the page you entered. What is the "No such special page" page? Is it Special:Nospecialpage? I would never know, because it would still display the message, telling me that it is not a valid special page. Ok, then I might as well look at the color of the link to know. But I still don't know, because it might be a red link like Special:Badtitle, where it is a red link although it is a valid target page. Is there an answer to this? --Gioguch (talk) 23:23, 11 October 2020 (UTC)[reply]

@Gioguch: Not everything is a page, and what you found is an example of something that isn't a page. See https://github.com/wikimedia/mediawiki/blob/6c257c213922017b3ebc6c9192653ec463b56c29/includes/specialpage/SpecialPageFactory.php#L748 for how that error gets displayed. Jackmcbarn (talk) 23:56, 11 October 2020 (UTC)[reply]

Browser jumps to edit box when editing

I have noticed this happening since probably late July. Whenever I try to preview a page that I am editing, my browser always jumps to the edit box. Since the default preferences for previews is above the edit box, my browser will jump to near the bottom of the page which will vary depending on how long the article you are previewing is. I noticed this also happens in other languages Wikipedia and on Commons, Meta but curiously, not on MW:MediaWiki. Does anyone know how to disable this? I want to start looking at the preview without concerning whether the page has done loading or not. It's also getting annoying when my Internet is slow so it becomes a surprise when I am reading while waiting for the page to complete but suddenly interrupted and get teleported to the bottom of the page. I have looked in preferences but couldn't find anything relevant other than setting so the preview appears below the edit box. But it will only make me jump to the top instead. RXerself (talk) 23:36, 11 October 2020 (UTC)[reply]

I have noticed that this only happens if I use the 2010 wikitext editor toolbar. It doesn't happen if I disabled it. It also doesn't happen when I'm previewing my common.js page. I have used the toolbar for years I don't remember it behaving this way. RXerself (talk) 07:30, 12 October 2020 (UTC)[reply]
Update: it looks like it only happen when I turn on the "Syntax highlighting" tool on the main namespace. If I turn it off it doesn't happen. If I am editing a subpage or a template it also doesn't happen even if the tool is active. RXerself (talk) 12:50, 12 October 2020 (UTC)[reply]

Massive memory use on Chrome/Mac?

I've had a weird issue where Wikipedia tabs take up a *massive* amount of memory (Chrome 85.0.4183.121/Mac 10.13.6). For example, a tab open to Sören Benn for the past few days currently has a Chrome memory footprint of 7.1 Gigabytes. Is there any known issue that might cause this to happen? power~enwiki (π, ν) 06:04, 12 October 2020 (UTC)[reply]

power~enwiki, try opening the same page in safemode. Could it be that one of your userscripts is causing the issue? —⁠andrybak (talk) 11:08, 12 October 2020 (UTC)[reply]

Archive script?

Relevant example - Wikicake with icing - is included for reference

Hi all, I was wondering if there is a script for autoarchiving? As in, a button that I can click that will add the relevant bot info and an archive box to an article's talk page?

Obviously this is not a critical problem but it is something that would probably prove to be the icing on my Wiki cake :) --Tom (LT) (talk) 07:13, 12 October 2020 (UTC)[reply]

There is help at Help:Archiving a talk page and examples at User:Lowercase sigmabot III/Archive HowTo — GhostInTheMachine talk to me 14:19, 12 October 2020 (UTC)[reply]

Parameter in thumb images to display attribution under caption?

Would it be possible with some templatestyles and /styles.css to introduce an additional parameter to the [[File:xyz.png|thumb|caption]] syntax to add an attribution field below the caption? Ideally, the attribution field would pull the author and license from commons, but otherwise manually defined as [[File:xyz.png|thumb|caption|attribution=Jane Smith, CC BY 4.0]]. I'm testing better automation of this template in my sandbox that avoids having to change between [[File:.. and {{fig|...! T.Shafee(Evo&Evo)talk 11:21, 12 October 2020 (UTC)[reply]

Clicking the image takes you to the file description page, which is sufficient for attribution. We do not want to clutter pages with attribution text, see MOS:CREDITS. --Redrose64 🌹 (talk) 12:44, 12 October 2020 (UTC)[reply]
... To answer the question, no, that will not work. Images are not extensible on wiki. --Izno (talk) 12:49, 12 October 2020 (UTC)[reply]

Linktrailing rules

Tracked in Phabricator
Task T47126

It is well known that the wikitext [[Foo]]bar produces a link that is six letters long: Foobar. But if the first letter after the closing braces is capitalised [[Foo]]Bar, the link is only three letters: FooBar. How long has this difference in behaviour existed? I first noticed it at Wikipedia talk:Criteria for speedy deletion#File PROD vs dfu (WP:F7d) in the post made at 13:26, 11 October 2020 (UTC) by Davidwr (talk · contribs) which contains the link WP:REFUNDED. --Redrose64 🌹 (talk) 12:41, 12 October 2020 (UTC)[reply]

@Redrose64: looks like it has been for at least 7 years according to phab:T47126 - notably it only matches [a-z] and fails on other things too like Foo2 , Fooä , etc. We wouldn't want it to trail on every character (especially not punctuation) - much more reading on it in the tracked phab ticket. — xaosflux Talk 14:24, 12 October 2020 (UTC)[reply]

WikiProject on open proxies discussion

I have started a discussion at Wikipedia talk:WikiProject on open proxies which may be of interest to some of the technical folks here. The thread is Wikipedia talk:WikiProject on open proxies#Reboot. GeneralNotability (talk) 14:05, 12 October 2020 (UTC)[reply]

"Top edits" oddity

If I try to look at my "Top edits" from this xtools page [1], I get No contributions found. But if try from "my" xtools page [2], things seems to be working. Something is wrong somewhere. Gråbergs Gråa Sång (talk) 14:17, 12 October 2020 (UTC)[reply]

Marking default gadgets

Hi all, I can't recall the old discussions on this - but wanted to get any input. This tech proposal is to include an indicator on gadgets that are enabled by default - so that if someone starts making preference changes they would more easily be able to revert to the defaults without having to erase all of their preferences. We've done this for a while on meta-wiki and you can see the inline indicators on your pref's page there. I've added a mock up to the default gadget for adding confirmation to mobile rollback, see in the second line of Special:Preferences#mw-prefsection-gadgets. Any feedback before updating other descriptions? Thanks! — xaosflux Talk 14:18, 12 October 2020 (UTC)[reply]

Do you mean this discussion back in July that resulted in Phab:T70689? — GhostInTheMachine talk to me 14:30, 12 October 2020 (UTC)[reply]
Great idea, but I think small text (with a trailing space) instead of superscript would be less ugly/confusing. Nardog (talk) 14:45, 12 October 2020 (UTC)[reply]
Meta shows a superscript "d" within the gadget description with a tooltip reading "Enabled for everyone by default". This does work and will help, but it is not very visible. Perhaps a larger label such as Default or Default or a big, bold blob  • of some sort? Just adding a plain text (Default) at the end of the description would also be good. — GhostInTheMachine talk to me 14:56, 12 October 2020 (UTC)[reply]
Or what about just an asterisk or a dagger or something at the end of each line, with a note at the top saying "* denotes gadgets enabled by default"? Nardog (talk) 15:02, 12 October 2020 (UTC)[reply]
Or even better, put these at the start of the line (with non-default lines padded) so that they display in a tidy column. That way, even a small mark will stand out. — GhostInTheMachine talk to me 16:31, 12 October 2020 (UTC)[reply]

Tech News: 2020-42

Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.

Problems

  • Because of the problems with the MediaWiki version two weeks ago last week's updates are also late. [3][4][5]

Changes later this week

  • Live previews didn't show the templates used in the preview if you just edited a section. This has now been fixed. You can also test CSS and JavaScript pages even if you have the live preview enabled. Previously this didn't work well. [6][7]
  • Recurrent item The new version of MediaWiki will be on test wikis and MediaWiki.org from 13 October. It will be on non-Wikipedia wikis and some Wikipedias from 14 October. It will be on all wikis from 15 October (calendar).

Future changes

Tech news prepared by Tech News writers and posted by bot • Contribute • Translate • Get help • Give feedback • Subscribe or unsubscribe.

15:23, 12 October 2020 (UTC)

Error message should not appear?

Not really a question about Wikipedia, but when you link to an invalid project (e.g. n:simple: for Simple English Wikinews), it will take you to Incubator. Using an unapproved language code (e.g. zxx and going to https://zxx.wikipedia.org will show an error message not produced by the WMF.) This should not be happening with existing codes, but testwiki:n: (Test Wikinews, not an existing project, takes you to a "This site can't be reached" error message not by WMF. IT should be taking you to incubator:Wn/testwiki or https://en.wikinews.org/wiki/Main_Page, but it's not and I believe this shouldn't be happening in local projects. Can a phab ticket help? --Gioguch (talk) 16:44, 12 October 2020 (UTC)[reply]

@Gioguch: This might be better suited for meta:Tech if it is unrelated to Wikipedia. (When would you write testwiki:n: instead of for example n:testwiki, and which resulting URL would you expect for each of them? I'm asking as your previous example n:simple: followed the "first language code, then site family" scheme. If I understand correctly... a clearer list of what you enter and which full URL you'd expect might help, to avoid misunderstandings.) --AKlapper (WMF) (talk) 16:59, 12 October 2020 (UTC)[reply]
Retrieved from "https://en.wikipedia.org/enwiki/w/index.php?title=Wikipedia:Village_pump_(technical)&oldid=983165914"