Policy | Technical | Proposals | Idea lab | WMF | Miscellaneous |
- Table of contents
- First discussion
- End of page
- New post
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.
Searching article text dumps
I would like to search articles in a way that I think is too complex for CirrusSearch and would time out (or at least be antisocial) on Search++. It isn't a job for Quarry because it needs article text. I've looked at PAWS but the examples there seem to read article text via API, which is feasible but not ideal (and I'd run it locally rather than add the complexity of PAWS). I could download an 86 GB dump but it would surely be far more efficient to search on a WMF machine which has access to that dump then download a few MB of results. Ideally, each job would run on one section of the dump, using a simple program. (I have a perl version ready.) However, I could just about get by with grep or equivalent. Is there such a facility anywhere?
In case this is an XY problem, what I'm trying to do is find certain links containing typos, such as [[Madonna|Madona]], which can result from an editor typing in a target badly then linking to the correctly spelled article via a VE dropdown. grep could handle single insertion (Maddonna), deletion (Madona), substitution (Medonna) or transposition (Maodnna), but anything more complex (such as Damerau–Levenshtein distance ≤ 2) really needs a proper programming language. Certes (talk) 23:39, 16 July 2022 (UTC)
- @Certes dumps are already available on Toolforge. And you can run whatever freely licensed/open source programs there. Legoktm (talk) 03:08, 17 July 2022 (UTC)
- Thanks, Legoktm. If anyone has successfully read article text from there then I'd love to see an example. I have a dev account (which I've barely used) and plenty of coding experience but would appreciate some hand-holding in gluing everything together. Certes (talk) 09:06, 17 July 2022 (UTC)
- @Certes: You can find the dumps you're looking for in
/public/dumps/public/enwiki/latest/
on any Toolforge server. You'll need a script to decompress the dump (you'll want to decompress on the fly) and parse the XML though, but it sounded like you already had something? Legoktm (talk) 04:11, 20 July 2022 (UTC)- Thanks; I'll look at getting a Toolforge project(?) started. I have a simple Perl program which I'm happy to release into the public domain. It takes standard input but could easily open a file or be run as
zcat some_dump | my_program
. I could add XML parsing if the appropriate Perl modules are available but may not need to bother as the regexps I'm looking for will in practice appear only in article text. I'll have to think about how to test this without running a process all day checking an entire mainspace dump, but I may be able to run it on a dump of a small namespace like Portal: first. Certes (talk) 11:13, 20 July 2022 (UTC)
- Thanks; I'll look at getting a Toolforge project(?) started. I have a simple Perl program which I'm happy to release into the public domain. It takes standard input but could easily open a file or be run as
- Some example methods for searching dumps: User:GreenC/software/search wikipedia. -- GreenC 04:30, 20 July 2022 (UTC)
- @Certes: You can find the dumps you're looking for in
- Thanks, Legoktm. If anyone has successfully read article text from there then I'd love to see an example. I have a dev account (which I've barely used) and plenty of coding experience but would appreciate some hand-holding in gluing everything together. Certes (talk) 09:06, 17 July 2022 (UTC)
- It's a pity
insource:/\[\[([^\|\]]+)\|\1~\]\]/
doesn't work (or does it?). — Guarapiranga ☎ 05:12, 20 July 2022 (UTC)- Good idea but a search for
insource:/(p)\1/ prefix:Apple
finds the pages with p1, not those with Apple.insource:/(p)$1/ prefix:Apple
finds nothing, presumably because there are no Apple articles containing literally p$1. Certes (talk) 11:13, 20 July 2022 (UTC)- Yeah, it seems Searching/Regex doesn't do capture groups (though I could find no mention of it). — Guarapiranga ☎ 22:39, 21 July 2022 (UTC)
- Good idea but a search for
Virtual keyboard behavior
When I am editing with the source editor, on an iPad (an old one, iOS 10.3.3) in desktop mode, the virtual pop-up keyboard "return" key maps to the normal carriage return as expected when typing in the edit window. But when I move to the edit summary input box, this key becomes "Go" and maps to "Publish changes". The problem for me is that I often hit this key by mistake when trying to hit the adjacent backspace key, leaving mangled and incomprehensible edit summaries that I can't fix. I would prefer only publishing my changes by closing the virtual keyboard and explicitly selecting the publish changes button. Can this be changed? MB 06:58, 18 July 2022 (UTC)
- Add the following to your common.js file:
// Prevent accidental submits
$(document).ready(function(){
$('input#wpSummary').keypress(function(e){
if(e.which==13) e.preventDefault();
});
});
- On any device, not just iPad, you'll no longer be able to accidentally submit with the enter/return/go key, and you'll just use the Publish changes button. I find it just as useful on desktop. MANdARAX XAЯAbИAM 09:24, 18 July 2022 (UTC)
- Thanks, I will try this when I get home to my desktop PC with a real keyboard, there is no way I am going to be able to edit my .js on this tablet. MB 20:26, 18 July 2022 (UTC)
It worked! Thank you very much indeed, it's a feature that I've been dreaming of for years. And let me remind you all that "your common.js" might as well be your global.js at Meta, so as to extend this to all the wikis you['ll ever] edit. (diff)
Could a bot assist with table column manipulation?
Courtesy link: WT:WikiProject Countries#Assistance requested with SYNTH issue in country demographic articles
Is there a bot that could assist with table column operations in dozens or hundreds of articles? I reported a SYNTH problem at this discussion involving what appeared to be a dozen articles, where the fix involves primarily dropping one column from a table on each article having the issue, plus a couple of other concomitant column tweaks. I can do that and have been plodding through, but what I thought was a dozen articles has ballooned to 165 now, and I don't think that's the end of it. It's too many to handle individually. Is there a batch operation that could assist with this? Thanks, Mathglot (talk) 07:17, 18 July 2022 (UTC)
- @Mathglot, I'd suggest posting this at WP:AWBREQ or, if there are >500 pages, WP:BOTREQ. If you've written some regex to do this, it would be helpful it you could post that as well. ― Qwerfjkltalk 08:27, 19 July 2022 (UTC)
- @Qwerfjkl: Thanks; currently it's up to 165. I probably can't use AWB, as the regex often has to be tweaked because of all the variation in styling among table cell definitions in the various articles. My latest is
s!(\|{1,2}\s*rowspan="\d"\s*)?(\|{1,2}){{(:?Flag\|)?[^|]+?}}\s*.(\|[\-}])!|-!gm
, and each version gets a little better at handling variation until there's some new variant I haven't seen before, so it sort of keeps evolving. Which is easy enough for me to tweak on the fly, but I guess would be tougher in AWB (dunno; haven't used it). So I suppose I'm stuck with my hand-rolled, ever-expanding regex. Thanks again, Mathglot (talk) 10:08, 20 July 2022 (UTC)- @Mathglot, I've had similar problems with Qwerfjkl (bot) 8 (code). Good luck. ― Qwerfjkltalk 11:09, 20 July 2022 (UTC)
- @Qwerfjkl: Thanks; currently it's up to 165. I probably can't use AWB, as the regex often has to be tweaked because of all the variation in styling among table cell definitions in the various articles. My latest is
CORS not allowing access to other projects
$.ajax( { url: 'https://de.wikipedia.org'+mw.config.get('wgArticlePath').replace('$1','Special:GadgetUsage?useskin=vector'), cache: false, type: 'GET' } );
I'm not asking for account data, this request can be made without cookies as far as I'm concerned.
It results in a CORS error. So I searched and found many solutions that don't work. In many questions about it it's unclear which options are for clients and which are for servers. I stumbled upon [1] which basically says "use this proxy". The stupid thing is: that actually works. It's obviously infinitely less secure and less reliable, but it works. *facepalm* There has got to be a better solution. — Alexis Jazz (talk or ping me) 21:11, 18 July 2022 (UTC)
- I found which makes the request without complaining, but no obvious way to get the page content. — Alexis Jazz (talk or ping me) 22:10, 18 July 2022 (UTC)
fetch('https://de.wikipedia.org'+mw.config.get('wgArticlePath').replace('$1','Special:GadgetUsage?useskin=vector'),{mode:'no-cors'}).then(response => {console.log(response);});
- Oh for fucks sake: https://stackoverflow.com/questions/43262121/trying-to-use-fetch-and-pass-in-mode-no-cors.
Long story short: in this case, evil people can easily get around CORS by using a proxy and good actors are forced to use a proxy as well because this is pretty dumb. What's the bloody point in blocking public resources? The actual request is still made, so if my goal was to send sensitive data toevil-incorporated.com
I still could. And if I need to retrieveevil-incorporated.com/mook.js
I still could. *facepalm* — Alexis Jazz (talk or ping me) 22:52, 18 July 2022 (UTC) - There is no better solution. This limitation of the web platform isn't to block you from retrieving whatever, it's to block you from retrieving it using someone else's device. As a simple example, if anyone on the internet could fetch https://de.wikipedia.org/wiki/Spezial:GadgetUsage from your browser, then they could extract your username from the result and associate it with your IP address. There are omre complicated scenarios, e.g. someone could access an intranet website that's only available from specific IP addresses. That's why using a proxy is the accepted workaround – then it's the proxy doing the retrieving from their own device and with their credentials, not with yours. Matma Rex talk 23:18, 18 July 2022 (UTC)
- (I'd recommend setting up your own proxy on Toolforge though, using someone else's proxy is indeed not exactly a good idea.) Matma Rex talk 23:20, 18 July 2022 (UTC)
- Matma Rex, thanks. Extraction of account details seems like it should be a non-issue: there should be a way to make the request without cookies. For the intranet, that would be a valid point, but an intranet site (which Wikipedia is not!) should specifically specify that its content should only be accessible by accepted domains. No, wait actually: if the data on the intranet site is sensitive in any way, it should be locked behind a login.
Also, using ToolForge as a proxy may not be allowed in this case. ToolForge rules are confusing with that. — Alexis Jazz (talk or ping me) 23:40, 18 July 2022 (UTC)- Even revealing the existence of an intranet site by allowing you to retrieve the login page is a privacy leak. Anyway, you're of course right that this could be done in a better way, but that is the platform we have to develop for and it has become this way for historical reasons; I thought you'd appreciate the explanation. There are sites on the internet (and definitely on intranets) that pre-date the introduction of CORS, and it has been specced this way to be compatible with sites that do not know of it.
- I don't see how that wouldn't be allowed on Toolforge, but what do I know.
- (Maybe the real solution to your problem would be to write a patch for the Gadgets extension that would make the gadget usage data available in the API somewhere. It seems like it could be added to action=query&list=gadgets easily enough. I'd be happy to review it if you cared to write it.) Matma Rex talk 02:00, 19 July 2022 (UTC)
- Matma Rex, I do appreciate the explanation, but it's still oddly designed and when "use a proxy" is the correct answer something has gone horribly wrong somewhere along the way. I could appreciate if intranet sites would specify that browsers should pretend they are unreachable when trying to access them from a foreign domain. I could also appreciate browsers doing that when they detect an internet site is trying to access something in a local range. But neither should be standard.
And say you know about some intranet site and you know how to hack it. Trick an employee into visiting your blog and you can make requests towards that intranet site! Never getting a response certainly complicates things, but the fact you can make requests at all should be worrying enough. Thinking about that, I'm starting to worry about routers.
Getting the usage data through the API would certainly be very nice. I'm pretty much completely unfamiliar with MediaWiki though and I haven't done any PHP in ages. I do wonder: if the API can be accessed from other domains, why can't other pages be accessed? If security is an issue, isn't there some render-as-anon option that could make CORS more lax? — Alexis Jazz (talk or ping me) 02:51, 19 July 2022 (UTC)- Consider if you were logged into your banking site and then visited a site with some malicious Javascript that made a request to access private information from your bank web site. To prevent it from going through, the cross-origin request has to be blocked by default. A proxy would be unable to access your info, as it would not be logged in as you. isaacl (talk) 03:46, 19 July 2022 (UTC)
- Isaacl, a request without cookies or other session data (compare opening a private window in your browser) would have the same effect for internet sites, wouldn't it? — Alexis Jazz (talk or ping me) 03:59, 19 July 2022 (UTC)
- Sure, making sure you're always logged out of your accounts would provide protection (and it's why this is a general security recommendation). In practice, users don't always log out, and don't browse every site with separate sessions restarted from scratch. Third-party Javascript code can be injected by third-party advertisers/trackers/discussion widget providers and so forth. Private browsing is, in a sense, overkill as not all cookies and other session data are problematic, but it's a reasonable default to wall off interactions between sites. In a similar manner, it's reasonable to block cross-site Javascript requests by default. isaacl (talk) 06:31, 19 July 2022 (UTC)
- I don't mean ask the user to log out, I mean more like
{mode:'no-cors'}
resulting in a request akin to how that request would be made if a private window was opened for it. That's what I initially expected from that option, but instead it just prevents you from getting any response at all. — Alexis Jazz (talk or ping me) 06:39, 19 July 2022 (UTC)- The native Javascript fetch API lets you specify that all credentials should be omitted, same origin credentials should be sent, or credentials should be sent in all requests (same origin or cross-origin). I don't know if the jQuery ajax API supports setting these options. isaacl (talk) 21:29, 19 July 2022 (UTC)
- I don't mean ask the user to log out, I mean more like
- Sure, making sure you're always logged out of your accounts would provide protection (and it's why this is a general security recommendation). In practice, users don't always log out, and don't browse every site with separate sessions restarted from scratch. Third-party Javascript code can be injected by third-party advertisers/trackers/discussion widget providers and so forth. Private browsing is, in a sense, overkill as not all cookies and other session data are problematic, but it's a reasonable default to wall off interactions between sites. In a similar manner, it's reasonable to block cross-site Javascript requests by default. isaacl (talk) 06:31, 19 July 2022 (UTC)
- Isaacl, a request without cookies or other session data (compare opening a private window in your browser) would have the same effect for internet sites, wouldn't it? — Alexis Jazz (talk or ping me) 03:59, 19 July 2022 (UTC)
- Yep, you definitely should worry about routers.
I do wonder: if the API can be accessed from other domains, why can't other pages be accessed? If security is an issue, isn't there some render-as-anon option that could make CORS more lax?
- I think in principle this would be possible, I don't know for sure why it's not allowed. I could guess that a) folks who worked on that wanted to encourage using the API instead of parsing the HTML output and b) it would require reviewing all of MediaWiki and extensions to make sure everything respects that parameter, and doesn't directly access global state like cookies. (There is a render-as-anon option in the API:
origin=*
, intended especially for anonymous CORS, for example: https://en.wikipedia.org/w/api.php?action=query&meta=userinfo&origin=*.) Matma Rex talk 10:47, 19 July 2022 (UTC)
- Consider if you were logged into your banking site and then visited a site with some malicious Javascript that made a request to access private information from your bank web site. To prevent it from going through, the cross-origin request has to be blocked by default. A proxy would be unable to access your info, as it would not be logged in as you. isaacl (talk) 03:46, 19 July 2022 (UTC)
- Matma Rex,
I don't see how that wouldn't be allowed on Toolforge, but what do I know.
wikitech:Wikitech:Cloud Services Terms of use: "Using Wikimedia Cloud Services as a network proxy: Do not use Wikimedia Cloud Services servers or projects to proxy or relay traffic for other servers. Examples of such activities include running Tor nodes, peer-to-peer network services, or VPNs to other networks. In other words, all network connections must originate from or terminate at Wikimedia Cloud Services."
Not so sure what I'd need would be allowed by that. Wikipedia isn't part of Wikimedia Cloud Services I think. And it certainly would be forbidden for any domain outside Wikimedia.
I thought about some script to create citations or similar things for stuff not supported by Citoid. Like importing ISBNs or something, processed by the client. CORS generally prevents that and ToolForge prohibits running a proxy to get the data from www.isbnsite.example to the user.
mw:Wikimedia Labs/Agreement to disclosure of personally identifiable information: "and agree your IP address will be made publicly available and not be treated as confidential". Okay, hard pass. — Alexis Jazz (talk or ping me) 05:09, 19 July 2022 (UTC)- A page that is labelled as a draft with talk page comments from 8 years ago? I don't think you have much to worry about there. Izno (talk) 05:32, 19 July 2022 (UTC)
- Izno, it's linked from wikitech:Portal:Toolforge/Quickstart though. — Alexis Jazz (talk or ping me) 06:25, 19 July 2022 (UTC)
- Huh, okay, I guess that does technically prohibit doing this, although I don't think that's the intended spirit of the rule. I'd ask for a clarification and I'd expect that doing what I suggested would be allowed. But I guess this is a moot point given SD0001's better solution below. Matma Rex talk 10:57, 19 July 2022 (UTC)
- I don't think it's even technically against the rules. For example, there used to be a tool that was really just a wrapper around Phabricator's API so it could be used in gadgets without needing an API key and to bypass CORS. Legoktm (talk) 19:14, 19 July 2022 (UTC)
- I agree with Izno that the random mw.o page you found is outdated, but to be clear, anyone who SSHs into Toolforge will have their IP made available to anyone else who is logged into the machine (run
w -i
). There's nothing stopping you from using a VPN, etc. to mask your real IP from Toolforge. Legoktm (talk) 19:16, 19 July 2022 (UTC)
- A page that is labelled as a draft with talk page comments from 8 years ago? I don't think you have much to worry about there. Izno (talk) 05:32, 19 July 2022 (UTC)
- Matma Rex, I do appreciate the explanation, but it's still oddly designed and when "use a proxy" is the correct answer something has gone horribly wrong somewhere along the way. I could appreciate if intranet sites would specify that browsers should pretend they are unreachable when trying to access them from a foreign domain. I could also appreciate browsers doing that when they detect an internet site is trying to access something in a local range. But neither should be standard.
- @Alexis Jazz Have you tried using mw.ForeignApi? — MusikAnimal talk 00:17, 19 July 2022 (UTC)
- MusikAnimal, how would I get Special:GadgetUsage (an HTML page) through ForeignApi? The data from GadgetUsage isn't available through the API afaik. — Alexis Jazz (talk or ping me) 00:50, 19 July 2022 (UTC)
- It's available through the API, like all other query pages. https://en.wikipedia.org/w/api.php?action=query&list=querypage&qppage=GadgetUsage&qplimit=max – SD0001 (talk) 10:42, 19 July 2022 (UTC)
- Thanks, I didn't even know this is possible! Matma Rex talk 10:59, 19 July 2022 (UTC)
- This API is missing the information about whether the gadget is default or not though, so you have to guess whether "0" means no users or all users… But I realized you can cross-reference this with the data in https://en.wikipedia.org/w/api.php?action=query&list=gadgets, which marks
default
gadgets. Together with that I think this matches the Special:GadgetUsage functionality. Matma Rex talk 11:03, 19 July 2022 (UTC) - Thanks, I'll use that+mw.ForeignApi! — Alexis Jazz (talk or ping me) 16:07, 19 July 2022 (UTC)
- It's available through the API, like all other query pages. https://en.wikipedia.org/w/api.php?action=query&list=querypage&qppage=GadgetUsage&qplimit=max – SD0001 (talk) 10:42, 19 July 2022 (UTC)
- MusikAnimal, how would I get Special:GadgetUsage (an HTML page) through ForeignApi? The data from GadgetUsage isn't available through the API afaik. — Alexis Jazz (talk or ping me) 00:50, 19 July 2022 (UTC)
- I suppose the simplest solution is to add the
origin
parameter to index.php the same way we have it on api.php. 0xDeadbeef 04:06, 19 July 2022 (UTC)
Template transcluding categories and SD
{{Get short description}} transcludes the target page to find the short description, but it also seems to transclude the categories and short description. Trying wrapping it in <nowiki>...</nowiki>
, or using {{#invoke:Page|getContent|expand}} doesn't seem to work either. ― Qwerfjkltalk 08:18, 19 July 2022 (UTC)
Template:Infobox television Lua errors on uncreated pages
When adding {{Infobox television}} to a new page on desktop and clicking "Show preview", I see two Lua errors. Firstly, at the top of the prose part of the screen: "Lua error in Module:Infobox_television at line 106: bad argument #1 to 'find' (string expected, got nil)." Secondly, at the top of the infobox: "Lua error in Module:Infobox_television at line 284: bad argument #1 to 'match' (string expected, got nil)." I don't see this error once the page has been created, either in the article proper or the preview screen.
I believe the issue is that the Lua is failing to grab the name of the page that the infobox is on, as the title at the top of the infobox does not display in the preview. I'm using the editing preference "Show previews without reloading the page", in case that's relevant. I note that I'm not the only one with this issue. It's just the TV infobox that's causing the issue: the page title is shown correctly and there are no Lua errors with {{Infobox film}}, for instance. — Bilorv (talk) 11:02, 19 July 2022 (UTC)
- I couldn't replicate it when testing it on Etetetetete. I however added nil checks to the lines the error message pointed to. In general though, post issues at the most relevent location. This isn't it. Gonnym (talk) 11:23, 19 July 2022 (UTC)
How can I get the editor with the citation tool and the editor with strike through, etc
Also, although I have "always given the source editor", I now default to visual editor with the ability to switch. How I make these messes is beyond me. Sorry folks. I do a lot of sourcing, so really need to be able to cite easily. Doug Weller talk 11:53, 19 July 2022 (UTC)
- And I now have a 🖊 instead of the word edit. Doug Weller talk 12:39, 19 July 2022 (UTC)
- Using Vector 2022. I now have an arrow at the top to publish. Doug Weller talk 13:14, 19 July 2022 (UTC)
- @Doug Weller if you want avoid the visual editor most of the time, try selecting "Temporarily disable the visual editor while it is in beta" in Special:Preferences#mw-prefsection-editing. — xaosflux Talk 13:33, 19 July 2022 (UTC)
- @Xaosflux That worked! Although now I have to switch between Wiked and whatever it is that shows as this. Easy enough to do though. Doug Weller talk 13:59, 19 July 2022 (UTC)
- And solved my inability to preview problem. Thanks so much. I'm dying of the heat here and this was a real timewaster, glad I had help, as always, from User:Xaosflux. Doug Weller talk 14:38, 19 July 2022 (UTC)
- Alternatively, on this wiki, you can go to Special:Preferences#mw-prefsection-editing-editor, turn on access to the visual editor, and in the drop-down menu right under that item, choose an "Editing mode" to "Always give me the source editor". (I personally prefer "Show me both editor tabs".) Whatamidoing (WMF) (talk) 00:17, 23 July 2022 (UTC)
- @Doug Weller if you want avoid the visual editor most of the time, try selecting "Temporarily disable the visual editor while it is in beta" in Special:Preferences#mw-prefsection-editing. — xaosflux Talk 13:33, 19 July 2022 (UTC)
- Using Vector 2022. I now have an arrow at the top to publish. Doug Weller talk 13:14, 19 July 2022 (UTC)
Search categories for both article and talk space conditions
Is there a way to search for articles matching an AND condition for categories amongst both the talk page and the main article page? eg. something like
All: incategory: "High-importance psychology articles" incategory: "All Wikipedia articles in need of updating"
The first category in this example is on article talk pages, the second category is on article pages. I know this search is not empty because of this table. If there is a way to do this, I think it would be worthwhile to add to Help:Searching/Features, Help:Searching, Wikipedia:FAQ/Categories and possibly also Wikipedia:FAQ/Categorization. Thanks Darcyisverycute (talk) 12:15, 19 July 2022 (UTC)
- Cirrus search can't do this, but PetScan may be able to. It can look for a category on the article and a template on the talk page. However, although All Wikipedia articles in need of updating is associated with the {{Update}} template, I can't find a suitable template for High-importance psychology articles. So my suggestion doesn't quite work, but someone may be able to fix it. If not then you might have to resort to a Quarry query. Certes (talk) 12:31, 19 July 2022 (UTC)
Move
How to undone move? Eurohunter (talk) 17:49, 19 July 2022 (UTC)
- Start a discussion on the talk page, and contact the editor who moved the article. It is not a technical problem, but rather a contested move which should be discussed. Donald Albury 17:59, 19 July 2022 (UTC) Edited 18:01, 19 July 2022 (UTC)
- @Eurohunter Since the move does not appear to have prior discussion or consensus you can go to Wikipedia:Requested moves/Technical requests and file an entry in the section "Requests to revert undiscussed moves" 192.76.8.85 (talk) 19:18, 19 July 2022 (UTC)
- Also, any autoconfirmed editor can simply move the page back, as the newly created redirect to the original title has a single line in the page history (Wikipedia:Moving a page#Moving over a redirect). DanCherek (talk) 19:27, 19 July 2022 (UTC)
- @DanCherek: True. I never remember about it. @Donald Albury: Start a discussion? Contested move? His move is actually wrong (move without reason - look at article) so it should be reverted without question and then he can start discussion about it. I could understand if I was the editor who moved page. Eurohunter (talk) 20:25, 19 July 2022 (UTC)
- @Eurohunter, as this is VPT, the answer is just "move it again". Are you having a technical problem with this? What happens when you try? — xaosflux Talk 13:08, 20 July 2022 (UTC)
- @Xaosflux: Yes I have done it - just forgot about this option (I din't not use it recently). Eurohunter (talk) 14:00, 20 July 2022 (UTC)
- @Eurohunter, as this is VPT, the answer is just "move it again". Are you having a technical problem with this? What happens when you try? — xaosflux Talk 13:08, 20 July 2022 (UTC)
- @DanCherek: True. I never remember about it. @Donald Albury: Start a discussion? Contested move? His move is actually wrong (move without reason - look at article) so it should be reverted without question and then he can start discussion about it. I could understand if I was the editor who moved page. Eurohunter (talk) 20:25, 19 July 2022 (UTC)
- Also, any autoconfirmed editor can simply move the page back, as the newly created redirect to the original title has a single line in the page history (Wikipedia:Moving a page#Moving over a redirect). DanCherek (talk) 19:27, 19 July 2022 (UTC)
Speedy delete candidate not showing in category
File:DiGiovanni competing on MasterChef.jpg is tagged for speedy deletion as a copyright violation via Twinkle since July 13, 2022. I was surprised it hasn't been deleted yet and checked the categories and found that although the image has categories on the file description page, it does not appear in the respective categories. The key ones are Category:Candidates for speedy deletion as copyright violations, and Category:Candidates for speedy deletion. Purging did not clear the problem. Whpq (talk) 20:58, 19 July 2022 (UTC)
- Purging does not change categorization. It needed a null edit. – Jonesey95 (talk) 21:32, 19 July 2022 (UTC)
- I am little confused. What null edit where? I see image is now in the categories but I don't see any null edit in the category or image. What am I missing? Whpq (talk) 21:47, 19 July 2022 (UTC)
- If either of these situations apply:
- page P shows category C at the bottom, but P is not listed at the page for category C
- the page for category C lists P, but C is not shown in the categories box at the bottom of page P
- this indicates that the link tables are out of synch. The fix for both is the same: go to P and perform a WP:NULLEDIT. A WP:PURGE does not update link tables. --Redrose64 🌹 (talk) 22:39, 19 July 2022 (UTC)
- Thanks for the explanation. My confusion came from thinking that a null edit was the same thing as a dummy edit. Now I understand they are not the same thing. Whpq (talk) 23:18, 19 July 2022 (UTC)
- Yeah, we have a lot of jargon here. In my experience fixing weird little problems, a null edit is always your best bet. Click Edit, even on a section, then click Publish without changing anything. If that doesn't fix the problem, then you really do have a problem that needs investigating. – Jonesey95 (talk) 05:29, 20 July 2022 (UTC)
- Thanks for the explanation. My confusion came from thinking that a null edit was the same thing as a dummy edit. Now I understand they are not the same thing. Whpq (talk) 23:18, 19 July 2022 (UTC)
- If either of these situations apply:
- I am little confused. What null edit where? I see image is now in the categories but I don't see any null edit in the category or image. What am I missing? Whpq (talk) 21:47, 19 July 2022 (UTC)
gadget notice
Wikipedia:EditNoticesOnMobile will be launching as a default gadget. It is limited to mobile users who are using Minerva skin. A waved approach is being done, wave one is only admins. Should something break, it can be instantly disabled via MediaWiki:Gadgets-definition. Please report any issues to Wikipedia talk:EditNoticesOnMobile. Best regards, — xaosflux Talk 23:04, 19 July 2022 (UTC)
Shouldn't deprecated ref/note templates simply be regexed out for ref tags
Shouldn't deprecated {{ref}}
/{{note}}
templates simply be regexed out for <ref>...</ref>
...
... like so: \{\{ref\|([^}]+)}}(.*?)\*?\s*\{\{note\|\1}}(.*?)\n
→ <ref name="$1">$3</ref>$2
— Guarapiranga ☎ 03:37, 20 July 2022 (UTC)
- Not like that, probably. Your regex would have to be a lot more sophisticated to account for all of the possible parameters in {{ref}}. I clicked on three articles at random that contained {{ref}}. One of them had a broken {{ref}} template that I had to replace with a real ref found in the article, so the regex definitely would not have fixed it. The next one, Jell-O, would probably not be compatible with your regex. The next one, Johnny Unitas, would probably also not be compatible. Regexes are powerful dark magic; they require both wisdom and patience to wield effectively. – Jonesey95 (talk) 05:38, 20 July 2022 (UTC)
- Also, "deprecated" means "don't make new instances", it doesn't mean "eliminate every existing occurrence". --Redrose64 🌹 (talk) 19:33, 20 July 2022 (UTC)
- Right. Guess I should've said couldn't... — Guarapiranga ☎ 08:05, 22 July 2022 (UTC)
Getting rid of the revision slider
How can I permanently turn off the revision slider when I go to any and every article's History? I can't find the option in my Preferences. — Cheers, Steelpillow (Talk) 07:08, 20 July 2022 (UTC)
- Put this in your CSS. —Cryptic 07:49, 20 July 2022 (UTC)
- Thanks. Works a treat. Bit impolite for the developers not to add a kill switch IMHO. — Cheers, Steelpillow (Talk) 09:55, 20 July 2022 (UTC)
- You shouldn't need the
!important
annotation. --Redrose64 🌹 (talk) 19:34, 20 July 2022 (UTC)
- I don't see a slider on article histories, but I do see one on diffs. There should be a preference to turn it off at Special:Preferences#mw-prefsection-rendering, the checkbox labeled "Don't show the revision slider". Anomie⚔ 11:57, 20 July 2022 (UTC)
- Every time we add a preference, it slows down page load times for everybody, on all pages. Since all that is needed is one very simple CSS rule (a class selector and one declaration) that any registered user can add to their personal CSS, it doesn't make sense to create a whole new pref. --Redrose64 🌹 (talk) 19:39, 20 July 2022 (UTC)
- The pref exists, as described by Anomie. Nthep (talk) 19:48, 20 July 2022 (UTC)
- Also RedRose's point isn't entirely accurate. Preferences only really slow us down if the preference would effect how wikitext renders. It doesn't matter much outside of that. There is however a substantial additional cognitive load that each added preference brings (proven in part by the fact that ppl apparently were apparently unable to find this preference). —TheDJ (talk • contribs) 09:47, 21 July 2022 (UTC)
- Every time we add a preference, it slows down page load times for everybody, on all pages. Since all that is needed is one very simple CSS rule (a class selector and one declaration) that any registered user can add to their personal CSS, it doesn't make sense to create a whole new pref. --Redrose64 🌹 (talk) 19:39, 20 July 2022 (UTC)
Pink paragraph
I assume this is related to User:Headbomb/unreliable script that I have installed. However, I ain't seen a whole paragraph turn pink. At HMS Terror (1813)#Legacy the paragraph starting "In July 2013, an anonymous miniaturist began reconstructing" is pink. Anyone know why? CambridgeBayWeather, Uqaqtuq (talk), Huliva 08:36, 20 July 2022 (UTC)
- @CambridgeBayWeather that userscript asks that questions about it be left at User talk:Headbomb/unreliable. — xaosflux Talk 09:52, 20 July 2022 (UTC)
- Yes but I don't know if the pink paragraph is caused by it or something else. CambridgeBayWeather, Uqaqtuq (talk), Huliva 10:18, 20 July 2022 (UTC)
- It is caused by that script. I don't know the script. PrimeHunter (talk) 10:58, 20 July 2022 (UTC)
- The easy way to tell if you suspect a userscript is causing you trouble is to simply turn it off (in this case via User:CambridgeBayWeather/common.js). You can always turn it back on again later. — xaosflux Talk 13:04, 20 July 2022 (UTC)
- Yes but I don't know if the pink paragraph is caused by it or something else. CambridgeBayWeather, Uqaqtuq (talk), Huliva 10:18, 20 July 2022 (UTC)
- The text
buildingterror.blogspot.com
. The script runs in more places than just references to catch issues like in works cited lists. Izno (talk) 16:55, 20 July 2022 (UTC)
Edit summary box - turn off enter to save changes?
If you cursor is in the edit summary box when you press enter, the edit finalises. I have, too often, caught the enter key when reaching for the backspace and saved a change I'm part way through. Is there a user setting to turn off this behaviour? (More than happy to be told it's a browser setting, rather than a WP setting, as long as I can turn it off) Little pob (talk) 13:47, 20 July 2022 (UTC)
- See #Virtual keyboard behavior above. MANdARAX XAЯAbИAM 13:55, 20 July 2022 (UTC)
- Your "common.js" file is: Special:MyPage/common.js — xaosflux Talk 13:58, 20 July 2022 (UTC)
- I hadn't included the link above because that user already had a common.js file, but it's useful for others (now, and later for people looking through the archives), so I added it there too. MANdARAX XAЯAbИAM 14:10, 20 July 2022 (UTC)
- Thank you, both! Little pob (talk) 14:27, 20 July 2022 (UTC)
- I hadn't included the link above because that user already had a common.js file, but it's useful for others (now, and later for people looking through the archives), so I added it there too. MANdARAX XAЯAbИAM 14:10, 20 July 2022 (UTC)
- Your "common.js" file is: Special:MyPage/common.js — xaosflux Talk 13:58, 20 July 2022 (UTC)
Why didn't this rename get logged properly?
Based on some hints (like the user talk page being redirected), I guessed that NDigital had been renamed to Niloybro, but couldn't find the log entry by searching for either name. Eventually I found this log entry by Nihonjoe but there's no performer and no target. The raw database record is here. What's going on? -- RoySmith (talk) 14:35, 20 July 2022 (UTC)
- @RoySmith it seems to work for me (with or without the User: prefix example 2), and the meta log is here: meta:Special:Redirect/logid/48837681. What is not working for you? — xaosflux Talk 14:41, 20 July 2022 (UTC)
- I've noticed that, sometimes, the logs take a few minutes (or longer) to be properly updated by whatever system does that during a rename. I'm not sure the reason why. ···日本穣 · 投稿 · Talk to Nihonjoe · Join WP Japan! 17:17, 20 July 2022 (UTC)
ParserFunction error
I'm completely useless when it comes to template coding. Why is the "North America" entry in List of continents and continental subregions by population returning an error message? This seems to be the result of Anderjef's updating of {{UN population}} on the 18th, but I can't figure out what the problem is. The string "{{formatnum:{{UN_Population|North America}}}}" (in the "List of continents ..." page) returns the correct value of 595,783,465, but the string "{{#expr:{{formatnum:{{UN_Population|North America}}|R}}/1e6 round 0}} million" returns an error message, even though the same string for South America, "{{#expr:{{formatnum:{{UN_Population|South America}}|R}}/1e6 round 0}} million", returns "434 million". This has me stumped. Deor (talk) 15:24, 20 July 2022 (UTC)
- I suspect that Editor Anderjef broke
{{UN population}}
with this edit (search for 'North America' in the right diff column).{{commas}}
wraps the value in<span>...</span>
tags and that confuses the{{#expr:}}
parser function. - —Trappist the monk (talk) 15:52, 20 July 2022 (UTC)
- @Trappist the monk: So I should just delete the
{{Commas}}
template there (and presumably in the "Channel Islands" entry immediately below)? Deor (talk) 17:42, 20 July 2022 (UTC)- I pinged Editor Anderjef so that that editor can explain why the edit was done as it was done. I don't know what the
<!--For backwards compatibility with 2019 version. July 2022-->
comments really mean so I don't know why Editor Anderjef used{{commas}}
instead of writing:| North America={{formatnum:{{#expr: 44092085+177050287+374641093}}}}
- (or doing it some other way)
- Were I you, I would hold off doing anything until Editor Anderjef responds.
- —Trappist the monk (talk) 17:56, 20 July 2022 (UTC)
- I'm not super experienced in template editing, so my mistake. I used {{commas}} to add commas to the numerical output to match all the rest (it passed the non-comprehensive testcases). The backwards compatibility stuff was for entries that had been listed in the 2019 UN data, but are not present in the 2022 data—I see reasons for keeping the North America and Channel Islands (and other such) entries if they were used somewhere in the past, but they don't literally appear in the data set (at least anymore) making them disingenuous to the source imo.
- I'd just replace the {{commas}} with
formatnum:
if that's expected to resolve the problem. Sorry! Anderjef (talk) 20:17, 20 July 2022 (UTC)- @Trappist the monk and @Deor Anderjef (talk) 20:18, 20 July 2022 (UTC)
- @Anderjef: As I said, I'm clueless when it comes to template coding, so don't know whether that would work. As I said in my original post above, the string in List of continents and continental subregions by population is "{{#expr:{{formatnum:{{UN_Population|North America}}|R}}/1e6 round 0}} million"; is introducing another
formatnum:
in there going to solve the problem? Part of the problem is that I can't tell whether{{UN_Population|North America}}
is used anywhere else on Wikipedia, although the "List of continents ..." is the only one showing up in Category:ParserFunction errors. (There doesn't seem to be a problem with the Channel Islands that I can see;{{UN_Population|Channel Islands}}
is used in the infobox of Channel Islands and displays the numeral correctly, with a comma, since it's not nested in other code.) Deor (talk) 20:46, 20 July 2022 (UTC)
- @Anderjef: As I said, I'm clueless when it comes to template coding, so don't know whether that would work. As I said in my original post above, the string in List of continents and continental subregions by population is "{{#expr:{{formatnum:{{UN_Population|North America}}|R}}/1e6 round 0}} million"; is introducing another
- @Trappist the monk and @Deor Anderjef (talk) 20:18, 20 July 2022 (UTC)
- Oh, and pings don't work when added after the fact as you did here.
- —Trappist the monk (talk) 17:59, 20 July 2022 (UTC)
- I pinged Editor Anderjef so that that editor can explain why the edit was done as it was done. I don't know what the
- @Trappist the monk: So I should just delete the
- My bug report with reproduction + screenshot is at: https://phabricator.wikimedia.org/T313187
- There I was told that I should report this here, as this is a CSS/configuration issue which must be solved on the specific WikiMedia server instance which is the Wikipedia (I use EN and DE).
- I hope that after the fix one can browse the Wikipedia with the Vector Skin's header on top and sticky headers stacking nicely below that. Thanks PutzfetzenORG (talk) 15:25, 20 July 2022 (UTC)
- @PutzfetzenORG:, I'm assuming you are referring to the testing and development gadget "
Make headers of tables display as long as the table is in view, i.e. "sticky" (requires Chrome v91+, Firefox v59+, or Safari)
", correct? Gadgets in that section are not well supported. The original author for that one can be contacted at User talk:TheDJ. If this is preventing you from viewing or editing pages right now, please disable it. If anyone has specific fixes to propose for it, please drop an edit request at MediaWiki talk:Gadget-StickyTableHeaders.css. — xaosflux Talk 15:34, 20 July 2022 (UTC)
Logos
How to re-create logos? More here. Eurohunter (talk) 17:10, 20 July 2022 (UTC)
- Eurohunter, The logos were created using Inkscape. It can be used to convert a PNG/JPG to SVG, but tracing simple images by hand would also work. 0xDeadbeef 17:43, 20 July 2022 (UTC)
- @0xDeadbeef: Any instruction? Eurohunter (talk) 18:59, 20 July 2022 (UTC)
- @Eurohunter Wikipedia:Graphics Lab/Illustration workshop --Ahecht (TALK
PAGE) 20:00, 20 July 2022 (UTC)- @Ahecht: I mean how I can do this myself. Eurohunter (talk) 21:29, 20 July 2022 (UTC)
- @Eurohunter Wikipedia:Graphics Lab/Illustration workshop --Ahecht (TALK
- @0xDeadbeef: Any instruction? Eurohunter (talk) 18:59, 20 July 2022 (UTC)
- Eurohunter:
- 1. Look up what font-face was used for the logo.
- 2. Compare that to m:SVG fonts, which is a list of fonts you can use in an SVG, all other fonts will render incorectly on WMF servers.
- 3. Embed the raster image (File - Import, options: embed, from file, any rendering)
- 4. Decide if the font exists at the previously mentioned "SVG fonts" list or not, if it does 4a, else 4b.
- 4a. Use the font tool and the font you found, align it with the raster image and find the correct font size.
- 4b. Trace the image by using the line tool (its the next item below the spiral in the toolbar). Alternatively download the font needed, use the font tool and then do path - object to path (this latter method is easier).
- 5. Remove the raster layer.
- 6. Save the image and upload, make sure to look at the thumbnail created before the upload and check if it is ok, if not, fix any errors it may have. Any questions?--Snævar (talk) 07:23, 21 July 2022 (UTC)
- @Snævar: I can't see this font on the list. Do you know what is this font? If you type "basshunter" "logo" "font" in Google Images there is "Hacker sklep komputerowy" and they has the same font but there is no name of font. "A" is similar to Comaro Font but letters "H" and "B" are slighty different and there are two different "S" letters so it had to be edited. There is also Babylon Industrial but it's not this too. Eurohunter (talk) 14:56, 21 July 2022 (UTC)
- No. Actually, most SVG text logos on commons are drawn. It is worth mentioning the SVG font list regardless, since it de-courages new SVG editors to make SVG files with unsupported fonts, I have seen a few of those.--Snævar (talk) 16:08, 21 July 2022 (UTC)
- @Snævar: It's all black when entered with Bezier. If I add straight line for B with Bezier then the second part is always curved. I need two straight parts then third curved line. How to control it? Eurohunter (talk) 16:51, 21 July 2022 (UTC)
- When you click on a bezier curve, there are little boxes on the line. When you click on that there is a line going through the box. The further the line goes away from the box, the bigger the curvature, and vice-versa. Also, you can angle the line and decide where it curves by doing that. Drag the boxes so there is one on a straight bit, and another one on a curvy bit, so it knows where the curve is. Snævar (talk) 19:09, 21 July 2022 (UTC)
- @Snævar: How I can undo one move instead of whole few steps? Eurohunter (talk) 22:05, 21 July 2022 (UTC)
- When you click on a bezier curve, there are little boxes on the line. When you click on that there is a line going through the box. The further the line goes away from the box, the bigger the curvature, and vice-versa. Also, you can angle the line and decide where it curves by doing that. Drag the boxes so there is one on a straight bit, and another one on a curvy bit, so it knows where the curve is. Snævar (talk) 19:09, 21 July 2022 (UTC)
- @Snævar: It's all black when entered with Bezier. If I add straight line for B with Bezier then the second part is always curved. I need two straight parts then third curved line. How to control it? Eurohunter (talk) 16:51, 21 July 2022 (UTC)
- No. Actually, most SVG text logos on commons are drawn. It is worth mentioning the SVG font list regardless, since it de-courages new SVG editors to make SVG files with unsupported fonts, I have seen a few of those.--Snævar (talk) 16:08, 21 July 2022 (UTC)
- @Snævar: I can't see this font on the list. Do you know what is this font? If you type "basshunter" "logo" "font" in Google Images there is "Hacker sklep komputerowy" and they has the same font but there is no name of font. "A" is similar to Comaro Font but letters "H" and "B" are slighty different and there are two different "S" letters so it had to be edited. There is also Babylon Industrial but it's not this too. Eurohunter (talk) 14:56, 21 July 2022 (UTC)
Coordinates in Vector 2022
Hello, on the Desktop Improvements/Vector 2022 talk page on mediawiki.org, there's sort of a "bug report", but we at the WMF can't fix it. It's about something that's totally up to the local communities - in this case, the English Wikipedia community.
It's about coordinates. As you can see on this screenshot, these are too close to the FA and page protection icons. @Jdlrobson has shared two ways of fixing the issue. I think we can say now that we strongly recommend using the indicator tag, so Jon's Option 1. It requires changes to Module:Coordinates.
@TheDJ, Sdkb, and Xaosflux: FYI.
Thank you! SGrabarczuk (WMF) (talk) 22:59, 20 July 2022 (UTC)
- @SGrabarczuk (WMF) Here is an example of a live overlapping page. Note, this template is used on over 1 million pages - so changes will need to be carefully considered and tested. Also, I think we'll need some editor to decide from a presentation viewpoint how this should be best laid out for our readers (assuming there certainly would be push back of "get that big language thing off of the top, it just sends readers away from our project!"). I actually sort of like coordinates where it is there - it is "content" after all; maybe the meta-indicators could move (perhaps to the left of the language bar??). — xaosflux Talk 23:44, 20 July 2022 (UTC)
- I replied on Phabricator, but I'll say again here that, as I've advocated elsewhere, I think the best place for the good/featured article icons is directly after the title, as that will make it far more prominent, which is needed for information literacy.
- For the protection icons, I think that should be associated with the edit button, since the main question for most people related to protection is "can I, with my current permissions, edit this page"? Minerva Neue handles this quite well, where if you can edit the page, you get an edit button, and if not, it's grayed out or locked or something and explains why. There's a bit of value also for just letting readers know the protection level so that they can understand how vulnerable the article they're reading may be to vandalism, but that's secondary and I don't think requires a very prominent notifier. {{u|Sdkb}} talk 01:28, 21 July 2022 (UTC)
- Having core handle icons for protection is phab:T12347. Izno (talk) 02:54, 21 July 2022 (UTC)
Authored By bzimport Jun 23 2007, 5:24 PM
. Ugh. Will there ever be any technical issue where we don't spend 20× as much energy flagging it as it'd take to actually solve? {{u|Sdkb}} talk 04:40, 21 July 2022 (UTC)- One option to consider is making coordinates a first class citizen in skins. @TheDJ: I made this POC to demonstrate one option which could be enabled on a skin basis. You could display:none the current coordinates in that skin to buy more time. Let me know if you feel this approach has any legs and is worth pursuing further: https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GeoData/+/816009 Jdlrobson (talk) 16:11, 21 July 2022 (UTC)
- Having core handle icons for protection is phab:T12347. Izno (talk) 02:54, 21 July 2022 (UTC)
- This is in progress, but it's hard when very long standing conventions have to be adapted. There are all kinds of hacks and overrides in place and multiple variants of templates that get affected once we start messing with the coordinates id (of which Module:Coordinates is the most prominent and most visible user) AND we have to keep into account all the skins, not just the one. Because of that, everyone is just afraid to break things and that causes stagnation in making progress in implementing the solution.
- Last but no least of all, even if we turn it into an indicator (which we really should), that doesn't mean the position and layout of the options makes sense to people. This has been conveniently ignored by the team so far, delegating the making of a new design to the communities, but the old design and positioning was a VERY old convention and is deeply settled in people's customs. Convincing people of a new layout in such a situation is challenging. —TheDJ (talk • contribs) 09:38, 21 July 2022 (UTC)
- The "layout" question is something to deal with indeed. SGrabarczuk (WMF) and maybe @OVasileva (WMF): - how much of this reading layout is going to be driven hard centrally, and how much is it lassie-faire for communities? Seems like we're already in the what tech changes to make stage - but the UX/UI layout regarding presentation of page meta-content isn't stable yet? I suppose the hard-line from mediawiki software side is: it works fine, stop putting these things on pages if you don't want them to break -- but our editors have found that this sort of content is well received with our readers. — xaosflux Talk 14:05, 21 July 2022 (UTC)
- @Xaosflux, I don't quite understand the last sentence. Could you rephrase? SGrabarczuk (WMF) (talk) 14:31, 21 July 2022 (UTC)
- @SGrabarczuk (WMF) in addition to the traditional "content" seen on pages, especially in encyclopedia articles, we have found it useful to inform readers and potential editors about meta-content, such as if we consider a page a quality page (FA, GA, etc), if we have applied restrictions to editing (protection status). In the most commons layouts (monobook, then vector). As there wasn't a designed section in mediawiki software for this meta-content display - we decided to place it above the content area. With the new vector-2022, the location we would use this for is now consumed with a new UI element (the language selector). Before we go about trying to make a "local" workaround for this, I'm asking if there is any design work from the reading team or other in the new skin team about how to present such indicators/meta-content in general for any project using mediawiki - and if so so, how firm current choices are — xaosflux Talk 14:44, 21 July 2022 (UTC)
- p.s. things like 'coordinates' are actual factual information about the subject of an article, so I don't really see them as 'meta-content'. — xaosflux Talk 14:45, 21 July 2022 (UTC)
- OK, ok. I've written a few GA's myself, so I totally understand the underlying issue. But then again, using the indicator tag for the coordinates wouldn't be local and wouldn't be a workaround. It'd be an application of the MediaWiki standard. As for the design work you mention, we did take a look, but that would be beyond Desktop Improvements 1.0, so we didn't pursue that path. SGrabarczuk (WMF) (talk) 14:56, 21 July 2022 (UTC)
- @SGrabarczuk (WMF) thanks, can you point to any "indicators" design layout/documentation pages that were updated for vector-2022? (Is mw:Help:Page status indicators the most current?) That page doesn't have any vector-2022 updates on it, but does use the "outside of the main content" design guide. — xaosflux Talk 15:03, 21 July 2022 (UTC)
- But it is a change from the current design. You force us to make a change and then say "well we didn't think about how it should be done, you figure it out". Now suddenly that makes the template editor the fixer, the designer and the wall for ppl to shout at because they don't like it. Can you see why that doesn't make it the highest thing on my and others 'lets relax with some wikipedia editing' evening ? ;) —TheDJ (talk • contribs) 15:03, 21 July 2022 (UTC)
- @SGrabarczuk (WMF) / @Jdlrobson - follow up on question about using "indicator" as suggested, it seems that the indicator design is also now "inside the content area" in vector 2022; compare the icon in the top right on vector vs vector-2022 on testwiki; which doesn't follow the notes at mw:Help:Page status indicators. Is there a vector-2022 "outside the content area" indicator section still? — xaosflux Talk 12:28, 22 July 2022 (UTC)
- @SGrabarczuk (WMF) in addition to the traditional "content" seen on pages, especially in encyclopedia articles, we have found it useful to inform readers and potential editors about meta-content, such as if we consider a page a quality page (FA, GA, etc), if we have applied restrictions to editing (protection status). In the most commons layouts (monobook, then vector). As there wasn't a designed section in mediawiki software for this meta-content display - we decided to place it above the content area. With the new vector-2022, the location we would use this for is now consumed with a new UI element (the language selector). Before we go about trying to make a "local" workaround for this, I'm asking if there is any design work from the reading team or other in the new skin team about how to present such indicators/meta-content in general for any project using mediawiki - and if so so, how firm current choices are — xaosflux Talk 14:44, 21 July 2022 (UTC)
- @Xaosflux, I don't quite understand the last sentence. Could you rephrase? SGrabarczuk (WMF) (talk) 14:31, 21 July 2022 (UTC)
- @TheDJ, you write "There are all kinds of hacks and overrides in place and multiple variants of templates that get affected" - and would this all be true if the indicator tag was used? Could you give some examples perhaps? SGrabarczuk (WMF) (talk) 14:43, 21 July 2022 (UTC)
- Template:Sky reused the coordinates id until recently, but also Module:Attached KML does (and probably a few more). These need to be dealt with as well. Then there is Template:Coordinates itself, which' output gets read by other modules (this is the infamous coordinsert hack). Then all the skins, which have different positioning and we need to put CSS in place that can work with both the before and after situation. All this requires hours of work and review and testing and moving slowly. And even if you can find a few hours to work on this it's not the most fun and once you are done, you might have to deal with the fallout (ppl yelling at you). —TheDJ (talk • contribs) 14:59, 21 July 2022 (UTC)
- I removed the instances Special:Search found last night (query, removals are from 19:56 EDT, 23:56 UTC?). None of them should have been using #coordinates besides Attached KML and Sky.
- Otherwise, I agree totally with TheDJ in his commentary here, especially when the research for moving ULS is as specious as it is. "We moved something above the fold, will it increase attention?" Gosh, IDK, will it? The question asked should have included "what lives in this space and will that decrease that attention". Izno (talk) 15:21, 21 July 2022 (UTC)
As for the design work you mention, we did take a look, but that would be beyond Desktop Improvements 1.0, so we didn't pursue that path.
@SGrabarczuk (WMF), I'll quote from my comment at VPR, which I understand you intend to reply to at some point: "I don't understand that — you consider it in scope to push them out but not to care about where they're pushed to?" {{u|Sdkb}} talk 15:46, 21 July 2022 (UTC)
- Template:Sky reused the coordinates id until recently, but also Module:Attached KML does (and probably a few more). These need to be dealt with as well. Then there is Template:Coordinates itself, which' output gets read by other modules (this is the infamous coordinsert hack). Then all the skins, which have different positioning and we need to put CSS in place that can work with both the before and after situation. All this requires hours of work and review and testing and moving slowly. And even if you can find a few hours to work on this it's not the most fun and once you are done, you might have to deal with the fallout (ppl yelling at you). —TheDJ (talk • contribs) 14:59, 21 July 2022 (UTC)
- The "layout" question is something to deal with indeed. SGrabarczuk (WMF) and maybe @OVasileva (WMF): - how much of this reading layout is going to be driven hard centrally, and how much is it lassie-faire for communities? Seems like we're already in the what tech changes to make stage - but the UX/UI layout regarding presentation of page meta-content isn't stable yet? I suppose the hard-line from mediawiki software side is: it works fine, stop putting these things on pages if you don't want them to break -- but our editors have found that this sort of content is well received with our readers. — xaosflux Talk 14:05, 21 July 2022 (UTC)
Legend colors
Hi
I was looking through some legends and saw colors swapped. So I fixed it. Today, a user reverted the change. Am I in the right or the wrong? Or is it possible that we are both right? [2] Trying to decide whether this is a technical issue.
jps (talk) 01:45, 21 July 2022 (UTC)
- @ජපස: Actually, you were wrong. Hold means the same party kept the seat after election. Gain means that incumbent party lost the election, and the other party gained the seat. In this case lighter colors denote HOLD, darker colors denote GAIN. So, your edit was wrong. Also, this isn't a technical issue but it's alright that you're here. In the future, better place for election-related questions, would be Wikipedia talk:WikiProject Elections and Referendums. —CX Zoom[he/him] (let's talk • {C•X}) 09:58, 21 July 2022 (UTC)
- So this is a technical issue, I think, because in my rendering of the legend, the colors are reversed. See image I uploaded. Could this be due to dark mode? jps (talk) 10:31, 21 July 2022 (UTC)
- Oh I see. Yes, indeed it is because of the dark mode gadget. —CX Zoom[he/him] (let's talk • {C•X}) 10:45, 21 July 2022 (UTC)
- Do I need to file a bug report or something? jps (talk) 10:50, 21 July 2022 (UTC)
- Don't make any color edits without checking the result in the default light mode used by nearly everybody, and always say if a color-related post is about dark mode. When "Dark mode toggle" is enabled at Special:Preferences#mw-prefsection-gadgets you should have a Dark/Light mode link at the top of every page. The issue is discussed at Template talk:Legend#Dark mode compatibility. PrimeHunter (talk) 11:35, 21 July 2022 (UTC)
- Related past discussion: Wikipedia:Village pump (technical)/Archive 195#Response of visual elements on using Dark mode (gadget). —CX Zoom[he/him] (let's talk • {C•X}) 11:43, 21 July 2022 (UTC)
- I've just deployed the changes discussed. But more edge cases likely exist, as legend can be combined with almost any type of other content, so it's always a bit of a guess. —TheDJ (talk • contribs) 14:43, 21 July 2022 (UTC)
- It looks like the legend colors at 2002_United_States_elections are still swapped in dark mode. Is there some purge that needs to be done to fix that now? jps (talk) 16:14, 21 July 2022 (UTC)
- Don't make any color edits without checking the result in the default light mode used by nearly everybody, and always say if a color-related post is about dark mode. When "Dark mode toggle" is enabled at Special:Preferences#mw-prefsection-gadgets you should have a Dark/Light mode link at the top of every page. The issue is discussed at Template talk:Legend#Dark mode compatibility. PrimeHunter (talk) 11:35, 21 July 2022 (UTC)
- Do I need to file a bug report or something? jps (talk) 10:50, 21 July 2022 (UTC)
- Oh I see. Yes, indeed it is because of the dark mode gadget. —CX Zoom[he/him] (let's talk • {C•X}) 10:45, 21 July 2022 (UTC)
- So this is a technical issue, I think, because in my rendering of the legend, the colors are reversed. See image I uploaded. Could this be due to dark mode? jps (talk) 10:31, 21 July 2022 (UTC)
Thanks for the links. What was not clear to me was whether there was any way to implement the class mw-no-invert in the legend template. It seems that this is not an option, but I'm not sure. Is there a way to include this in "legend" as an optional feature? jps (talk) 11:52, 21 July 2022 (UTC)
Is there a need for a RFC to discuss splitting rights in Mediawiki and implement it?
Some background information: The zhwiki applied to enable the function of the recent patroling (mw:Manual:$wgUseRCPatrol) . I raised an objection, thinking that this would overlap with the function of the new page patroling (The new page patroling is a job with a designated user group, and The recent patroling is a voluntary work without a designated user group), and then some users who support enabling this function plan to split the "patrol" right from the function, with this task: phab:T308153.
And It has now been two months, and this task is still very deserted, and no one has discussed or reviewed the implemented code. So I have a question about this: Does this change to Mediawiki functionality require a RFC to deal with so that the developers of Mediawiki or Foundation can review and follow up?
Also, if a RFC is made to modify the functionality of Mediawiki, where should I apply? mw, or meta, or RFC?
Cwek (talk) 03:49, 21 July 2022 (UTC)
- RFCs don't really modify functionality. They provide the basis for proving that there is wide community desire for a particular thing. This says little about implementation cost, deploy cost, planning, long term maintainability etc, which is all up to either WMF or a volunteer. Reviewers also will not magically appear, they have to be actively sought out.
- However without an RFC showing the desire there might be little motivation to prioritise at all. —TheDJ (talk • contribs) 09:27, 21 July 2022 (UTC)
Template doc header
Well friends, I'm stumped again – isn't the first time and won't be the last. Seems that a recent change automated the application of {{Documentation subpage}} to not just template /doc pages but also to all the redirects to those pages. An example is Module:Infobox3cols/doc, which targets its associated template /doc page. My own humble opinion is that the doc header shouldn't be on redirects, it should only go on actual template documentation pages. Placing the doc header on all subpages named "/doc" gives an incorrect transclusion number for the "Documentation subpage" template. I think the interface page that transcludes the template is MediaWiki:Scribunto-doc-page-header, however I'm not certain of that. Just wondered if any of youse-much-more-savvy-than-I technical minds can figure a way to remove the doc header from all the /doc page redirects? P.I. Ellsworth , ed. put'r there 12:12, 21 July 2022 (UTC)
- The only think I can think of is to use a module for checking whether a page is a redirect or not. Module:Redirect should work (using its
isRedirect
feature). Then, we can put a test on the MediaWiki page to see if the page is a redirect, and hide the header if it is. --ais523 12:55, 21 July 2022 (UTC)- Thank you! that's what I was thinking, too. I just want to be sure we have the correct MW page before we put an edit request on its talk page. I'm not too strong when it comes to modules, so I'd need help with adding the
isRedirect
feature to the interface page. P.I. Ellsworth , ed. put'r there 13:07, 21 July 2022 (UTC)
- Thank you! that's what I was thinking, too. I just want to be sure we have the correct MW page before we put an edit request on its talk page. I'm not too strong when it comes to modules, so I'd need help with adding the
Here is another with auto-generated text, this time it asks editors to discuss the sandbox page. And of course redirects are not the place for discussion, not even talk-page redirects. I don't think these are being tested so as to eliminate usage on certain pages. They're just being "hopefully applied". I can identify with that; however, these messages on redirects should be erased. P.I. Ellsworth , ed. put'r there 19:45, 21 July 2022 (UTC)
- @Paine Ellsworth: I've now created {{redirect other}}, which (assuming that it works in MediaWiki:-space; I think it will, but MediaWiki:-space is weird sometimes) should make it easy to implement a change like this. (Note that the template will need to be fully-protected in order to be usable as part of the site interface.) --ais523 23:14, 21 July 2022 (UTC)
Template Editor needed. Serious neutrality violation with Template:Infobox_election
The software selects one image from the lead section of an article as the PageImage, prominently representing the article in various places. When this happens with Template:Infobox_election for an upcoming election, Wikipedia prominently promotes one candidate over the other. Not only does this directly and severely violate Neutrality, it undermines Wikipedia's reputation by appearing to take sides in the election.
The needed fix is to edit the template to add |class=notpageimage to each candidate image. That prevents the images from being selected as PageImage.
If you can confidently determine an election has concluded, the ideal result would be to have all losers tagged as |class=notpageimage and allow the winner to be the PageImage. However if a winner cannot be confidently determined within template code, the priority must be to ensure none of the candidate images are used. Alsee (talk) 19:26, 21 July 2022 (UTC)
- An image of one of the leading candidates seems better to me than no image. Should we also randomly rotate them in the infobox so nobody is always shown first in the article? This seems more a content issue than a technical matter. You could try to get consensus at Wikipedia talk:WikiProject Elections and Referendums or Template talk:Infobox election with the other linking to the discussion. If you get consensus but miss somebody who can implement it then it becomes a technical matter. I see there are already posts at Template talk:Infobox election but they don't discuss whether to do it at all. I would certainly start with that before deliberately disabling a significant MediaWiki feature which can help readers find the article they are looking for, and also has many other uses. PrimeHunter (talk) 21:48, 21 July 2022 (UTC)
User with problem with "Create Account"
Drypuglia has been asking me for help on my user page and talk page. After telling him to come here he added this question at the top of the FAQ page:
Cannot "Create Account" been trying everything for some time! Been active a long time but they tell me I don't exist-please help!Drypuglia. Drypugl... serius
Since he has an account I am not sure what the problem is. StarryGrandma (talk) 19:35, 21 July 2022 (UTC)
- Drypuglia, this is the place to chat with someone about your problem. People here have a lot more knowledge than I do. But we need more information. You have an account. Why do you need to create another one? When you click on Special:CreateAccount and enter a new user name, what happens? What is the exact error message you get? StarryGrandma (talk) 21:08, 21 July 2022 (UTC)
- Drypuglia: You have a working account here at the English Wikipedia at https://en.wikipedia.org. If you want an account somewhere else then please link it. If it's another domain where you are able to log in with the same username and password then your Wikipedia account also works there. You are logged in if your username is displayed at top of the page. If the name is red then you have not created a user page there and may see a message which confuses you but don't worry. It's optional to create a user page and your account works fine without it. You do have a user page User:Drypuglia here at https://en.wikipedia.org. PrimeHunter (talk) 21:21, 21 July 2022 (UTC)
- I have a user page. When I tried to write a post and upload a photo -Iwas blocked because I did not have an account! Drypuglia (talk) 21:51, 21 July 2022 (UTC)
- @Drypuglia: Please post a link to where you tried to write a post and upload a photo. If you got an error message then quote it exactly. PrimeHunter (talk) 22:09, 21 July 2022 (UTC)
- Probably tried to upload to our sister site Commons, but had not yet logged into Commons (same account, separate login) —TheDJ (talk • contribs) 08:13, 22 July 2022 (UTC)
- I have a user page. When I tried to write a post and upload a photo -Iwas blocked because I did not have an account! Drypuglia (talk) 21:51, 21 July 2022 (UTC)
- Drypuglia: You have a working account here at the English Wikipedia at https://en.wikipedia.org. If you want an account somewhere else then please link it. If it's another domain where you are able to log in with the same username and password then your Wikipedia account also works there. You are logged in if your username is displayed at top of the page. If the name is red then you have not created a user page there and may see a message which confuses you but don't worry. It's optional to create a user page and your account works fine without it. You do have a user page User:Drypuglia here at https://en.wikipedia.org. PrimeHunter (talk) 21:21, 21 July 2022 (UTC)
Searching Putler - Articles without term appear
I assume a template is causing this but when I do a search for "Putler", I get 347 results. 2 actually include the term. The other 345 show "Hey, Rise Up!" Madonna of Kyiv [ro] "Oi u luzi chervona kalyna" Patron "Putler" Saint Javelin "Slava Ukraini!" "Stefania" "Ukraine (song)" "Ukraine on" which is not in the articles or source of the sampling i checked.
insight? Slywriter (talk) 22:12, 21 July 2022 (UTC)
- @Slywriter It is one of the links in Template:2022 Russian invasion of Ukraine (in "Impact"), used on many pages. Search results include text generated by templates. Matma Rex talk 22:22, 21 July 2022 (UTC)
- (edit conflict) @Slywriter: "Putler" is in the "Impact" section of Template:2022 Russian invasion of Ukraine which is displayed at the bottom of many articles in the desktop version of the site. The mobile version omits such navigation templates but still includes the articles in the search result. I see you are currently using the mobile version so I can understand your confusion. insource:Putler searches for articles where the source text contains "Putler". PrimeHunter (talk) 22:28, 21 July 2022 (UTC)
Vector 2022 display overflowing issue caused by Template:Para
There seem to be an display overflowing issue with Template:Para when being used in replies, and when viewed on Vector 2022 skin. When viewed using Vector vs Vector 2022 in Template talk:Infobox musical artist. — Paper9oll (🔔 • 📝) 07:26, 22 July 2022 (UTC)
- Fixed Para, which made use of nowrap on potentially very long content, which is a nono. —TheDJ (talk • contribs) 08:07, 22 July 2022 (UTC)
New subpage behaviour?
I created a talk page subpage for an archive yesterday and instead of a blank page that I would simply add the 'AAN' template to I got a complicated screen inviting me to start a new discussion. This seems new to me, is there any way to bypass it and carry on as normal? I believe it is driven by the forward slash in the page name. Thanks. Nimbus (Cumulus nimbus floats by) 07:54, 22 July 2022 (UTC)
- Yes, you click the create link on that same page, or to disable permanently you go to Special:Preferences#mw-prefsection-editing-discussion and set the appropriate "quick topic adding option" —TheDJ (talk • contribs) 08:11, 22 July 2022 (UTC)
- Thanks very much, I've reset my preferences. Nimbus (Cumulus nimbus floats by) 08:27, 22 July 2022 (UTC)
DNG/Raw image upload support
From the linked phab this has been a long-standing request, is there anything that has changed to allow editors to upload DNG/NEF/etc files? My other thought was seeing if maybe we could make a tool to add the raw file as an entry inside of a JPEG file (with a corresponding tool to extract the raw file for someone interested). This is not ideal as it would add an additional payload to the JPEG file (obviously significantly increasing its size), but would at least be a way to assure that editors wishing to contribute un-edited/raw images have a way to do so that is on-wiki and can be maintained in the attribution history correctly. —Locke Cole • t • c 19:31, 22 July 2022 (UTC)
- Adding new file types is probably being held up by phab:T294484, where there isn't anyone who's actually maintaining the system that produces thumbnails and it's woefully out of date. --Ahecht (TALK
PAGE) 21:18, 22 July 2022 (UTC)- Honestly, it would be enough to at least be able to upload (and of course, download) the files. Thumbnails wouldn't really be necessary as the edited JPG/PNG/etc would presumably be available. We shouldn't be displaying raw files directly in articles/pages. Someone mentioned the patented state of DNG during objections a number of years ago, but as the format is now nearly 20 years old, those initial patents should be expiring (or possibly already expired if they were awarded prior to DNG being documented). —Locke Cole • t • c 20:49, 23 July 2022 (UTC)
- There is more information to this. There is an working group working solely on updating the thumbnailing system and they are half way there, see phab:project/profile/5851/ for that. The steward requests, which the aforementioned phab bug is, have been blocked by the lack of an CTO (Chief Product and Technology officer). In August there will be a new CTO at WMF, which then enables handling those again. That weather the thumbnails should be made or not is irrelevant, because it is the same system that changes from DNG to other raster formats as does the thumbnailing. Lastly, Brion mentioned that he would be updating documentation of how to add more file formats, once the thumbnail system update is done, so that will become easier. In short, there is hope for this bug, but nothing concrete can be said until after at least one month.--Snævar (talk) 23:15, 23 July 2022 (UTC)
- While the thumbor problem would be a major blocker to deploy, its not as if there is anything to deploy. Ive added the instructions to how to add support to that ticket a while ago, and anyone able to find the TIFF metadata ids that make a tiff file a DNG file could implement this and start contributing patches. Please remember that the foundation doesnt have dedicated multimedia engineers, all file support is primarily driven by volunteers at this moment in time. The best way to make things happen is by writing code. —TheDJ (talk • contribs) 10:41, 24 July 2022 (UTC)
- Are thumbnails required? Because I'd honestly settle just for having the ability to upload/download DNG files, which from the instructions you linked, should be as simple as just setting up the MIME types/extension..? —Locke Cole • t • c 22:40, 24 July 2022 (UTC)
- No, as described in https://www.mediawiki.org/wiki/Manual:Adding_support_for_new_filetypes —TheDJ (talk • contribs) 07:54, 25 July 2022 (UTC)
- Are thumbnails required? Because I'd honestly settle just for having the ability to upload/download DNG files, which from the instructions you linked, should be as simple as just setting up the MIME types/extension..? —Locke Cole • t • c 22:40, 24 July 2022 (UTC)
- Honestly, it would be enough to at least be able to upload (and of course, download) the files. Thumbnails wouldn't really be necessary as the edited JPG/PNG/etc would presumably be available. We shouldn't be displaying raw files directly in articles/pages. Someone mentioned the patented state of DNG during objections a number of years ago, but as the format is now nearly 20 years old, those initial patents should be expiring (or possibly already expired if they were awarded prior to DNG being documented). —Locke Cole • t • c 20:49, 23 July 2022 (UTC)
Correlating edits with page titles?
Are there any good tools for showing what title a page existed under at the time an edit was made? I'm looking at two users accused of socking. User A has edited a page in user B's sandbox. Normally, that sounds like socking. But, if the page was in mainspace when A made their edit, and later it was moved to B's sandbox, that's less suspicious. The data to figure this out certainly exists in the move log, but doing the correlation by hand is a pain. Has somebody already written this tool? -- RoySmith (talk) 15:46, 24 July 2022 (UTC)
- User:Nardog/MoveHistory: This tool lists all the page moves a page has gone through in a chronological tabular format. So it can help do what you're looking for. —CX Zoom[he/him] (let's talk • {C•X}) 19:38, 24 July 2022 (UTC)
No merge for alternate accounts?
Some years back, I was given a legitimate alternate account. I now see that accounts cannot be merged: Wikipedia:Changing username. Would somebody just confirm that I cannot merge my two legitimate accounts?
Thanks. `HG1 | Talk 17:07, 24 July 2022 (UTC)
- Confirmed. mw:Extension:UserMerge is not installed here: Special:Version. PrimeHunter (talk) 18:53, 24 July 2022 (UTC)
Ignoring quotation marks in sorting by name
For literary works in a sortable table it looks like class="wikitable sortable" first sorts those in quotation marks (short stories) and then novels in italics, so in effect two separate lists are formed. Is there a way to ignore quotation marks so that both short stories and novels would be sorted in a single alphabetical order? Brandmeistertalk 12:04, 25 July 2022 (UTC)
- See Help:Sorting#Specifying a sort key for a cell. There has to be code in each cell with different sorting. If most cells have a quotation mark then you can flip it and add a sort key with a quotation mark in the other cells. PrimeHunter (talk) 12:47, 25 July 2022 (UTC)
BLP notice
When {{WikiProject Biography}} is placed inside the {{WikiProject banner shell}} wrapper, the automatically genereated notice for living people does not appear when the wrapper is exapanded. Is this a known issue, and easily fixable? UnitedStatesian (talk) 17:48, 25 July 2022 (UTC)