Tech/Archives/2020

From Meta, a Wikimedia project coordination wiki
< Tech‎ | Archives
Jump to navigation Jump to search

somthing wrong in a template of Hakka wikipedia

The template hak:模板:Infobox officeholder looks ugly. Let's see the article hak:Chhai Yîn-vùn, why it looks quite different from en:Tsai Ing-wen? can someone fix it?--el caballero de los Leones (talk) 12:33, 12 September 2019 (UTC)

@唐吉訶德的侍從: Which exactly difference do you mean? The large width is caused by the {{nowrap}} in the birth_place parameter value. Ankry (talk) 14:15, 23 February 2020 (UTC)
Now it's fixed. Thank you very much.--el caballero de los Leones (talk) 04:16, 29 February 2020 (UTC)

Getting rate-limited (http 429) requesting thumbnails using mediawiki api

Hello. I am attempting to use the mediawiki api to make a simple page that loads thumbnails of every image from a given category page of wikimedia commons. It works fine for category pages with only a few images, like Commons:Category:Monogram_of_Albrecht_Altdorfer, which has 17 images. However, I get http 429 errors when attempting to load thumbnails from a category page with many images, like Commons:Category:Albrecht_Altdorfer, which has 164 images. Is there a more appropriate way to request many thumbnails using the api like this? I figured it should be possible since I would essentially be requesting the same data a normal user requests on loading a given category page. My initial api request is: https://commons.wikimedia.org/w/api.php?origin=*&action=query&redirects=1&format=json&generator=categorymembers&gcmtype=file&prop=imageinfo&iiprop=timestamp%7Curl%7Csize%7Cdimensions%7Cmime%7Cthumbmime%7Cmediatype&iiurlwidth=150&iiurlheight=150&gcmtitle=Category:Albrecht_Altdorfer, which if there are more than a few images returns a continue attribute, and subsequent requests are the same but with those continue values appended. The exact javascript I am using is below. Any help understanding this would be greatly appreciated!

var params = {
  action: "query",
  redirects: 1,
  format: "json",
  generator: "categorymembers",
  gcmtype: "file",
  gcmtitle: "Category:Albrecht_Altdorfer",
  prop: "imageinfo",
  iiprop: "timestamp|url|size|dimensions|mime|thumbmime|mediatype",
  iiurlwidth: 150,
  iiurlheight: 150
};

async function getResponse(params) {
  let url = "https://commons.wikimedia.org/w/api.php?origin=*";
  Object.keys(params).forEach(key => url += "&" + key + "=" + params[key]);
  const json = await fetch(url)
    .then(response => response.json());
  let pages = json.query.pages;
  if (json.continue) {
    for (let key in json.continue) {
      params[key] = json.continue[key];
    }
    let pages1 = await getResponse(params);
    pages = {...pages, ...pages1};
  }
  return pages;
}

Jmviz (talk) 00:39, 1 January 2020 (UTC)

(Personally) Opt Out of Partial Block

I have a strong belief that partial block is wrong and I do not wish to see/use them on my block interface. Given that it is being enabled on most wiki by next week (Community health initiative/Partial blocks#December 17), I would like to know if it is possible to hide the partial block stuff (leaving site-wide only). (Query NKohli & SPoore). — regards, Revi 01:31, 4 January 2020 (UTC)

Hello User:-revi, as you have noted, partial block will deploy on most wikis today (with exceptions of two large projects, Wikimedia Commons & English Wikipedia, whose contributors are currently discussing if and how to use partial block.) As you know, all admins and other functionaries are free to opt out of using block, deletion, protection, checkuser, etc. and their different levels or types of use. Being a volunteer, they can select the ways that they want to contribute to the Wikimedia movement. Unfortunately, opting out of doing a particular type of work does not normally allow for parts of interface to be hidden. And we did not design it this way for Special:Block when partial block was added. So, it will not be possible for you to hide partial blocks.
I also want to add that I have read your reasons for not wanting partial block to be implemented and respect you for speaking out. Even though it appears that more people support deploying partial block, I think it is important for all points of view to be heard. SPoore (WMF) Strategist, Community health initiative (talk) 22:46, 6 January 2020 (UTC)
Well, I can live with that, even though that will mean slight discomfort on mobile emergency-blocking (page length). Also, I am fine with the new tools being implemented — more flexibility for someone who wants it! I just don't want to be disturbed by the feature I will never use. :-p — regards, Revi 23:55, 7 January 2020 (UTC)

E-mail adress for Wikinews

Hello,

Is possible to create an e-mail adress for frwikinews : frwikinews@wikimedia.org ? It's necessary for social networks, but our actually e-mail is on OutLook, a Microsoft product, so we're don't independent.

Thanks, AirSThib (Flight attendant · Flights), the 08:58, 19 January 2020 (UTC).

@AirSThib: Technically it should be possible using OTRS, although I don’t know whether such use of OTRS is allowed. You can contact one of the OTRS admins listed there. —Tacsipacsi (talk) 16:31, 19 January 2020 (UTC)
I think you can request to add a new mailing list, here. --Gryllida 20:16, 19 January 2020 (UTC)
@Tacsipacsi and Gryllida: Where are the request, because it's we want ! AirSThib (Flight attendant · Flights), the 17:50, 20 January 2020 (UTC).
  1. If you need a shared inbox where you and other community members can collectively respond: OTRS is the right venue.
  2. If you need a messaging group where you and other community members can discuss your stuff: Mailing lists.
— regards, Revi 20:40, 20 January 2020 (UTC)
Also note that you are only allowed to have a personal inbox on (_AT_)wikimedia.org if you work for / provider service to the Wikimedia Foundation.
If you go by OTRS, I think the naming scheme is something like info-fr(_AT_)wikinews.org or wikinews-fr(_AT_)wikimedia.org sort of thing (I can't guarantee, I'm not responsible for granting OTRS addresses), and for mailing lists I think it's probably wikinews-fr(_AT_)lists.wikimedia.org under the current naming convention. — regards, Revi 20:48, 20 January 2020 (UTC)
@AirSThib:Do you want a public mailing list, or a private OTRS e-mail where only privileged helpers can read new messages from people who are looking for help? Please confirm. Thank you. --Gryllida 23:38, 20 January 2020 (UTC)
@Gryllida, Tacsipacsi, and -revi: We want an e-mail for social networks. What is the better ? AirSThib (Flight attendant · Flights), the 14:39, 25 January 2020 (UTC).
Public mailing list is a big no. I don’t want to imagine a password reminder archived in a publicly visible mailing list archive, accessible to everyone with an Internet access… —Tacsipacsi (talk) 14:47, 25 January 2020 (UTC)
@Gryllida, Tacsipacsi, and -revi: So what e-mail we should use ? AirSThib (Flight attendant · Flights), the 15:18, 1 February 2020 (UTC).
@-revi, Tacsipacsi, and Gryllida: So if I understand we must use an OTRS e-mail ? AirSThib (Flight attendant · Flights), le 14:36, 2 February 2020 (UTC).
@-revi, Tacsipacsi, and Gryllida: OTRS e-mail ? AirSThib (Flight attendant · Flights), le 11:32, 23 February 2020 (UTC).
We’ve provided you with options. This is not a kindergarten, so it’s your turn to decide and take action. —Tacsipacsi (talk) 13:40, 23 February 2020 (UTC)
@Tacsipacsi: I have re-read the talk and I choose OTRS. AirSThib (Flight attendant · Flights), le 13:44, 23 February 2020 (UTC).

English Wikinews already has an OTRS queue, linked here (en.wn), wikinews-en at wikimedia dot org. Does French Wikinews already have one? This page (fr.wn) does not say, and nor does this one (Meta). --Gryllida 01:27, 24 February 2020 (UTC)

Correction: there is wikinews OTRS queue for all sister projects. You could apply for access to that. @AirSThib see OTRS/Volunteering. Regards --Gryllida 01:38, 24 February 2020 (UTC)
@Gryllida: So it's info-fr@wikimedia.org, no ? AirSThib (Flight attendant · Flights), le 09:06, 25 February 2020 (UTC).
No, that's French Wikipedia. --Gryllida 09:17, 25 February 2020 (UTC)
@Gryllida: In n:fr:Wikinews:Contact, have wikinews@wikimedia.org. It's nice ? AirSThib (Flight attendant · Flights), le 09:23, 25 February 2020 (UTC).
Yes, it is correct. This is the 'Sister projects::Wikinews' OTRS queue. --Gryllida 09:29, 25 February 2020 (UTC)
@Gryllida: So only OTRS have access to e-mail address ? AirSThib (Flight attendant · Flights), le 17:27, 27 February 2020 (UTC).

Automaticly switch to mobile/desktop mode based on browser OS.

Switch link is "hidden" at the bottom of the page and it is used only if user is desparate to view right rendering. Why anyone on desktop would like to see mobile version? At least move switch link top of the page.

Hi 109.240.40.206
  • Have you seen other web sites where mobile version is readily available at the top of the page - if so please give examples?
  • Logged in users can also select the Timeless skin, like this, in preferences - it is pretty responsive and consistent.
Regards,--Gryllida 05:05, 13 February 2020 (UTC)

cronjob to download mysqldump and image folder

I have a server-side cronjob running that produces a regular mysqldump with date in the name. Now I need a script that makes a incremental backup of my image-folder and also downloads the dump. That would work with rsynch, curl and sshpass, but I wonder if I need to reinvent the wheel, as I guess 100s of other users have done the same. Can anyone point me to such a solution?


Kind regards

2A02:8109:9880:C4C:C1FF:87A3:FB21:83FD 19:48, 16 February 2020 (UTC)

This page is about Wikimedia wikis. It seems your query would be more appropriate at mediawiki.org
Why use sshpass? You control the server, so you should be able to use ssh keys instead.
A command like tar -C wiki.folder -cf "backups/today-differential.tar" --listed-incremental=uploads-metadata images should provide you with differential dumps of your image folder.
Platonides (talk) 20:07, 16 February 2020 (UTC)

Edit the preamble in articles of Wikipedia

Hello! How can I enable the ability to edit the preamble in Wikipedia articles? Please help me in this matter. Thanks! -- Adam-Yourist (talk) 20:59, 26 March 2020 (UTC)

Hi Adam-Yourist. Editing the intro or lead section of a wiki article can be a little confusing. The easiest way is to click the "edit" or "edit source" tab at the top of the page. Or there are user gadgets that add a dedicated link to the user interface. More info here: w:en:Help:Section#Editing before the first section. --MZMcBride (talk) 22:09, 26 March 2020 (UTC)
Thank you for the answer, MZMcBride! But, I have a problem in our Wikipedia in Ingush. Firstly, I do not have all the gadgets reflected in the settings (as it is for example here in the English Wikipedia). Secondly, we want the ability to edit the introductory section (preamble) of an article to always be enabled for all users (as it is, for example, in Russian Wikipedia, the "edit" tab is placed on the right side). Do you have information on how to do this? I will be very grateful for the help. -- Adam-Yourist (talk) 08:27, 27 March 2020 (UTC)
You can make it a gadget and enable it for everyone ([...|default]). I believe this is how other wikis do it. --Matěj Suchánek (talk) 10:43, 27 March 2020 (UTC)

Edittools at el.wiktionary

Is this the correct place to ask for help for https://el.wiktionary.org/wiki/MediaWiki:Edittools ? It does not function. The charinsert> does not work, and we have noone who can correct https://el.wiktionary.org/wiki/MediaWiki:Common.js If only we could copy some Phonetic characters at least! it would be terribly helpul. Thank you! Sarri.greek (talk) 13:46, 14 March 2020 (UTC)

Please, where is the place for Small Wiktionaries Help? Technical help? Sarri.greek (talk) 04:50, 30 March 2020 (UTC)

How system understand that 03/29/20 is 29 March?

Hi! When I translated uk:Template:Interactive COVID-19 maps/Cumulative cases I had a problem that system understands 03/20/20 as March 20 but in Ukrainian March will be "Березень". Please look at template in ukwiki and drag the circle. You will see that "Бер" was changed to "Mar". Where I can fix it? Thanks!--Dimon2712 (talk) 18:31, 29 March 2020 (UTC)

Edit by section

I cannot edit by section even though there's no coding of "NOEDITSECTION". Is this a technical problem or my device problem? Please do check it and reply to me ASAP. Thank you! CyberTroopers (talk) 08:58, 10 April 2020 (UTC)

@CyberTroopers: Hi, what page are you referring to? Majavah (talk) 09:04, 10 April 2020 (UTC)
@Majavah: All pages in Malay Wikipedia, English Wikipedia.. I opened it using my phone but there's no "pencil icon" at any section. Is there any configuration changes about it? CyberTroopers (talk) 09:08, 10 April 2020 (UTC)
@CyberTroopers: Thanks, the part about this being a mobile issue helped searching. This looks like to be this bug: task T249864. Majavah (talk) 09:16, 10 April 2020 (UTC)
@Majavah: Alright thank you. CyberTroopers (talk) 09:21, 10 April 2020 (UTC)

campaign-external-machine-translation

Hello, what is this new edit tag? Is it part of the Content translation tool or Visual editor or both? I can't find any documentation about it. Here's an example: [1]. -kyykaarme (talk) 18:39, 20 April 2020 (UTC)

It comes when you start translating a page you first translated via Google Translate. For example you open google translate page from enwiki to Finnish, then there is a header that says "contribute" (or something by local language, in Finnish it's nonsense "edistävät"), and when you click it you go to that special page on fiwiki where you can start the page either from scratch with VE or using the content translation tool with VE. Stryn (talk) 19:09, 20 April 2020 (UTC)
Thanks! I tried it, and it was, hm, interesting. To use Content translation you need to have a username, and to use Google translation you need to try it somewhere else than on enwiki. And if you're translating into Finnish, you first have to click a button that makes no sense. :) -kyykaarme (talk) 16:06, 24 April 2020 (UTC)

Large bot-generated lists

On Wikipedia, we can make bots that check articles (i.e. by catscan). Bots make a formated list and write it to the page not in the main namespace.

  • the result can be long, so it approaches one of the technical limits for page. For example, a page can be so large that it cannot be included in another page.
  • the result can be updated once a day, once every few days, or once a week.
  • this can be done for many wiki-projects or for monitoring the status of a specific list of articles

Only current data is needed and previous records are not needed. Often several lists are made on separate pages and they are included on one page - only the current lists are visible (similar to the special page of recent changes), and and the history does not need the history of those separate pages.

Hence a couple of questions:

  • How does the Wikimedia Foundation look at the fact that its resources are wasted not for articles, but to save unnecessary bot generated data? For example, a new version of an article is generated every day that is close to the maximum size or several small lists, and all this is done for several wiki projects every day forever. Or maybe we can freely create versions for hundreds of lists every day without hesitation? (on the one hand it's close to the “pillars of wikipedia” and to “what wikipedia is not”, on the other hand, if there was an explicitly declared permission or confidence that this does not create an unnecessary load, then we would deploy such lists more massively (more lists per wikiproject, more often updates, more statistics) and for more wiki-projects with fewer participants.)
  • Is it possible not to save old versions somehow? If this is beneficial for server maintenance and convenient for editors - can there be a namespace where old versions will not be saved or, for example, will be only the last 10, or it would be possible to switch the page to a new mode (Special:ChangeContentModel), or may not save marked pages to dumps - we can re-generate them, or, if it is profitable to maintain, can we make a list of pages / versions that can be permanently deleted after reaching community agreement on them, lists can even be written to another wiki (for example, data in a test wiki can be deleted) if they can be included in Wikipedia pages (non-main space), like files/datasets included from the commons-wiki.
  • Are there other ways to make it easier for servers to work with bot-lists? Are there bottlenecks (perhaps in the long run?) for servers when working with bot-lists?
  • Are there any guidelines, instructions, wishes from the WMF regarding such renewable bot-generated lists?
  • Can it somehow help, that sometimes old versions will be deleted for these pages? (with the current tool, versions are not deleted, but hidden)
  • Can creating and updating such lists in a separate (new?) namespace somehow help?

--Sunpriat (talk) 06:28, 7 May 2020 (UTC)

How could I get bigdata?

Hello! I need to get person name, date of birth, date of death, image url for each person in list of Category:Born_on_day_month for each day in a year (For each day about 1000 persons). When I tried to get a list by https://ru.wikipedia.org/w/api.php?action=query&format=xml&list=categorymembers&cmlimit=100000000&cmtitle=Category:родившиеся_1_января it`s all ok. But when I tried to get wiki id for each person by https://ru.wikipedia.org/w/api.php?format=xml&action=query&prop=pageprops&titles=Суворов,_Александр_Васильевич and then by https://www.wikidata.org/w/api.php?format=xml&action=wbgetclaims&entity=Q154232 it was already error after 5-10 requests. One request duration is about 0.5 s. I see, it`s very irrational way and hope to find better one. What is the best way for this purpose (maybe one big request for some subpurpose)? Thanks

@IoanGum: Hi (please sign your posts), what is the exact error? --AKlapper (WMF) (talk) 16:55, 7 May 2020 (UTC)

3 Questions

I feel that new MediaWiki makes the source editor harder to use.

  1. Where should I write this impression ?
  2. Where were the specifications for this revision considered ?
  3. Where can we write the request for the next revision ?

--HaussmannSaintLazare (talk) 19:14, 9 May 2020 (UTC)

@HaussmannSaintLazare: Hi, 1) You just wrote it here (though we do not know what your impression is based on). 2) What "revision" of what exactly? 3) What "request" and what "next revision"? I'm afraid this thread currently needs way more context. --AKlapper (WMF) (talk) 19:20, 9 May 2020 (UTC)

thema

thelw na kanw to profil otan kaneis anazitisi sto google me to onoma mou na fenaite i oikonoma mou i stadrodromia mou imeromenia genisis olo pws mporo na to kanw giati den katexo to wipedia katholou thelw na me prosotanolisis

Κάθε άρθρο στην Wikipedia, είτε στην αγγλική έκδοση ή την ελληνική ή οποιαδήποτε άλλη, θα πρέπει να είναι για θέμα που αξίζει να βρίσκεται σε εγκυκλοπαίδεια (δες w:el:Βικιπαίδεια:Εγκυκλοπαιδικότητα ή w:en:Wikipedia:Notability). Επιπλέον, θεωρείται σύγκρουση συμφερόντων το να γράψει κάποιος πληροφορίες σε άρθρο για τον εαυτό του, δες w:el:Βικιπαίδεια:Σύγκρουση_κινήτρων ή w:en:Wikipedia:Conflict_of_interest. Αν όμως ενδιαφέρεσαι να συμμετάσχεις στην επέκταση της Wikipedia γενικά, μπορείς να ρίξεις μια ματιά εδώ: w:en:Wikipedia:Contributing_to_Wikipedia. Καλή επιτυχία! -- ArielGlenn (talk) 13:42, 17 May 2020 (UTC)

New Page and subjects

How exactly do I start a new page? And what can the page be about? —The preceding unsigned comment was added by Normal Person exploring this world (talk)

Hello, see Wikimedia projects to see what kind of pages they contain. See mw:Help:Starting a new page on how to create a page. Nemo 15:45, 19 May 2020 (UTC)

Anti-vandalism bot

Hi, I'm sysop at sc.wiki, and I have been having a problem with an incredibly annoying vandal for a while. Since Sardinian has a few different written standards, our pages can be written using any of those, and the community decided that the standard used by others can't be changed unless a few specific conditions are respected, but this guy it's targeting the pages written using the sc:LSC standard and the logudorese one, marked with Template:LSC or with Template:Variant. Since I can't watch the changes 24/7, and that If I ban him he changes IP and just starts again every time, it's there a way to create a bot that reverses any change that deletes or modifies those templates (unless it deletes Template:Variant, but adds Template:LSC, that allows multiple standards) unless I approve it? Also, it's there any other tool that I can use to stop him?--L2212 (talk) 12:43, 7 May 2020 (UTC)

Hi L2212. Are you familiar with the AbuseFilter extension? It may fit your needs. --MZMcBride (talk) 18:56, 17 May 2020 (UTC)
Thank you very much MZMcBride, it looks exactly like what we needed! I have another question, though. Studying the manuals and looking at the rights of the users, it looks like the autoconfirmed and extendedconfirmed rights are given, on sc.wiki, differently to how they are given on en.wiki. Where can I look at our local settings, and if necessary, change them?--L2212 (talk) 16:22, 19 May 2020 (UTC)
Hi L2212. There is a reference copy of the Wikimedia wikis configuration available at <https://noc.wikimedia.org/conf/InitialiseSettings.php.txt>. You can file a Phabricator Maniphest task to request a wiki configuration change once there's local consensus. --MZMcBride (talk) 19:08, 19 May 2020 (UTC)
Ok, thank you very much again!--L2212 (talk) 22:58, 22 May 2020 (UTC)

What to use as client_id for OAuth2

I'm trying to make a react based client to be hosted gitlab.io (static hosting) with PKCE flow (more info here).

Documentation relating to OAuth2 is here: [2].

When I register an OAuth2 consumer at https://meta.wikimedia.org/wiki/Special:OAuthConsumerRegistration/list I get 3 values, labeled as:

  • "Client application key"
  • "Client application secret"
  • "Access token"

I have tried all 3 of the values I got as client_id and none work. I navigate to to https://meta.wikimedia.org/w/rest.php/oauth2/authorize?client_id=...&redirect_uri=...&response_type=code&scope=openid&state=,..&code_challenge=...&code_challenge_method=S256&response_mode=query

But this page tells me "Application Connection Error: Client authentication failed (e.g., unknown client, no client authentication included, or unsupported authentication method)". So I guess I'm doing something wrong, first step would be to verify that I am indeed using the correct thing for "client_id".

The documentation says to use "client token" as "client_id", but I dunno what that is, I guessed it is "Client application key" but as that does not work I guess it is not.

Any ideas on what I'm doing wrong?

I'm using hueshika000. There is a related phabricator issue here.

If someone has an example client that does "PKCE flow" that I can have a look at it will be great. Iwan Aucamp (talk) 13:17, 23 May 2020 (UTC)

Some CSS for Vector has been simplified

Hello! I'd like to make a double-check about a change that was announced in Tech/News/2020/21.

Over-qualified CSS selectors have been changed. div#p-personal, div#p-navigation, div#p-interaction, div#p-tb, div#p-lang, div#p-namespaces or div#p-variants are now all removed of the div qualifier, as in for example it is #p-personal, #p-navigation …. This is so the skins can use HTML5 elements. If your gadgets or user styles used them you will have to update them. This only impacts the Vector skin.

On this wiki, this impacted or still impacts the following pages:

How to proceed now? Just visit all these pages and remove div before these CSS selectors if it hasn't been removed so far. Thank you! SGrabarczuk (WMF) (talk) 11:25, 25 May 2020 (UTC)

My one cent of unhappiness that WMF did not give us enough (if any?) time to prepare for the change. — regards, Revi 12:00, 25 May 2020 (UTC)

MediaWiki forum

I am making a mediawiki stack exchange site. Anyone who wants to contribute can and should. 72.74.131.76 14:59, 23 May 2020 (UTC) The proposal is up at https://area51.stackexchange.com/proposals/124244/mediawiki 65.96.125.113 11:08, 26 May 2020 (UTC)

Starting new wikis

I was told this website allows you to create wikis. How could I do that?

Please sign with four tildes. You need a web server. Install the software and read the docs. 65.96.125.113 11:11, 26 May 2020 (UTC)
You probably mean Fandom. Ruslik (talk) 20:54, 26 May 2020 (UTC)
Miraheze lets you create wikis. --Rob Kam (talk) 07:20, 27 May 2020 (UTC)

Restricting the use of Content translation tool

If I have understood correctly, the English Wikipedia has restricted the use of the Content translation tool by 1) requiring 500 edits (extended confirmed user group) before the editor is able to publish a translated article straight to mainspace and 2) disabling machine translation completely. Does any other Wikipedia have similar or other restrictions with the tool? And would it be technically possible to limit the total use of the tool to a certain user group? So not only limiting the publishing feature of the tool like enwiki does, but hiding or otherwise restricting the tool completely from editors who are not part of the user group? -kyykaarme (talk) 22:22, 28 May 2020 (UTC)

You can find the current configuration in InitialiseSettings.php. I far as I can see only enwiki has set such limitations. Ruslik (talk) 18:29, 2 June 2020 (UTC)

Lua help to small wiki

Dear Sirs, hello from el.wiktionary. I know nothing about Lua. Need to make a Module for some 40 placenames for my wiktionary.
I have trouble with a simple version of it: I tried all kinds of combinations to get me 'error' and I cannot, I cannot do it.
Could someone help? I know is is insultingly easy... But there is noone around anymore who would know how to do it. Thank you
--Sarri.greek (talk) 00:26, 4 June 2020 (UTC)
Done.
Small little things, that would make the difference for small wikis: 'how to' little modules. Like, applications with the commands in a non-latin languages (it took me one week to solve). Like, how to make a Category from a word. Like, How to link a word. Like...
bad argument #1 to 'sub' (string expected, got table).

local export = {}
function export.test(text)
	return mw.getContentLanguage():ucfirst(mw.ustring.sub(text, 1, 1)) .. mw.ustring.sub(text, 2)
end
return export

Sarri.greek (talk) 23:38, 5 June 2020 (UTC)

Unwatching deleted pages

I don't know if this is the right place for this, but can we devise a more convenient way to on mobile unwatch pages that have been deleted? Currently, we have to click on the red link (which sends us to the editor interface), remove ?action=edit or #/editor/all from the URL, and then finally press the star to unwatch. This process is often repeated if you've been watching multiple doomed pages. Glades12 (talk) 05:22, 7 June 2020 (UTC)

Unreachable username

Just like unreachable lowercase usernames in the early days of Wikipedia, the username ";" is now unreachable. See [[3]]. The user has made several edits, but their user pages give bad title errors unless it is reached from a redirect. Can someone look into this? Cheers! CrazyBoy826 (talk) 19:14, 12 June 2020 (UTC)

The username ':' and ': harping.mike ::' are also unreachable, although they are only global accounts and do not have any attached local ones. CrazyBoy826 (talk) 19:19, 12 June 2020 (UTC)

Bot

Hello!

   G'👉 Ǵ
   g '👉 ǵ
   N'👉 Ń
   n '👉 ń
   A'👉 Á
   a '👉 á
   U'👉 Ú
   u'👉 ú
   O'👉 Ó 
   o👉 ó 

Please change these letters to the new alphabetical article titles on Karakalpak Wikipedia and all articles according to these rules.(BEGJIGIT (talk) 12:11, 13 June 2020 (UTC))

Why and who? Stryn (talk) 15:25, 13 June 2020 (UTC)

IDEA FOR MAKING WIKIPEDIA MORE TRUSTWORTHY

If you could generate a X-ray mode that highlights in shades the text in an article that is more recent, less recent, and colorize sections of the article that fluctuate, where edits and removals are occurring frequently. Or the ability to identify the content that hasn't been changed significantly, excusing minor edits. These visualization aids could be toggled to give the reader better understanding of the text's progress. Some of these I know would require the collection of metadata from the DIFF function (knowing something of the UNIX command that is often used to discern edits between files) over time to reduce the immediate cost of applying the filters. The effects should be accessible as special tags in a clone of the article, not to appear with the article, to avoid confusion, through access of the xml/html display, so that third party software such as web browser extensions could do advanced and much snazzier effects on the articles visualization.

I think by adding this, you will win over the segment of the population that is easily thrown off when they realize that a wiki is crowdsourced. This segment of the population don't understand that wikis retain most all edits, and if a visualization mode was added and could be shown, it may be the necessary proof to help those recognize how wikis are much better than published encyclopedias in the past.

PS- More could be done to make this better such as identifying stuff in the article that was added by unconfirmed users, edits by recent members, edits by established users, edits by users with infrequent involvement, edits by users who have been away for x amount of time in the past (to identify possible ghost accounts).

Keep in mind all of this should be just a mode that can flipped on like a light switch, possibly added to the content after it is displayed, and resides on an invisible layer, then when switched on is brought to the front. It could be initially implemented as just a set of alternate renderings that are displayed in place of the basic rendering. But I'm sure with some thought it could be made reliable and manageable.


Extra chatter : Better visual feedback is what separates the best software from the better, really bad software is that which relies heavily on things like nested menus, is modal (offering different functions in multiple modes, and failing to make the mode apparent by periphery of the interface, such as colorization or the addition of special borders, an icon is not enough of a cue) and offers almost no visual or aural feedback. You'd get these kinds of concerns in a course HCI (Human Computer Interaction/Interface Design). The reason Apple products don't need manuals is the features are well thought and conservatively introduced, but constant change of the interface may actually hurt even their products adoption especially if users have become accustomed to the original interface, interface changes often throw off fragile users, and can lose them easily, even make them cry, as a child temporarily separated from its mother in a grocery store cart. These users are often elderly, and in some cultures, elderly are influencers, if they fail to use a product, the family may never adopt the product. Keep in mind, in south of the US, where I am, Texas specifically, many of the less educated "country hicks" tend to learn everything and adopt and share the opinions of their elders, and they are the most vocal, and most wrong when it comes to necessary change. By making them as a target audience, you get everyone else who is more adaptable and flexible. The design shouldn't pander to one particular crowd, but the worst users should be the best testers. Also using metadata, it is possible to dumb down interfaces and substitute things for purposes of accessibility. Another idea I have is to have a group that focuses on making articles accessible to commuters and offering special parsers that users can adjust to customize how words are said in articles, when converted to audio by voice synthesizers. There is software for Windows called TextAloud that offers a Regular-Expression controlled pre-processor filter for users to customize the text of an article to be vocalized by a particular voice. At one time I had constructed a dictionary of these Regular expressions to read unedited wikipedia articles from mouse-select-copied text from the raw web page, and the end result was much more understandable than without use of this. If such software was builtin to wikis, using either serverside mods, or cloud-services (HINT, HINT, MAKE MONEY, HINT, HINT) that could be offered to augment the experience of hearing the articles, then such extras could be offered as a premium service, even separate from the wikis public offering.

Ideas never hurt. I tell people the difference between philosophy and religion is whether you can live life by it, if you believe it. The bible is at least philosophy. Ideas. And Ideas should not be ignored. —The preceding unsigned comment was added by 47.185.89.90 (talk) 12:50, 25 June 2020 (UTC)

Fileinfo error on MW 1.34

I'm receiving a MW 1.34 internal error; required components "fileinfo"

FileInfo folder containing Fileinfo.php, FileInfoi18nmagic.php are under the wiki/ extensions location and the php_fileinfo.dll extension in php.ini is enabled. I'm not sure what else I'm missing.


MS Visual C++ 2019

PHP 7.2.31

MySQL 5.7.17

@DOSSOD: This page is for Wikimedia wikis only (see above) but no Wikimedia wiki runs MW 1.34. Please see mw:Project:Support_desk and please also sign your posts (see above). Thanks! --AKlapper (WMF) (talk) 00:30, 27 June 2020 (UTC)

New features to WikiMedia.

Can you add some new features to WikiMedia? —The preceding unsigned comment was added by 2A02:C7F:DC4D:D200:C95C:428F:DBF8:2A4F (talk)

I can't. Stryn (talk) 17:37, 27 June 2020 (UTC)

Authenticating a request from a user script to a toolforge tool

I would like to authenticate a request from a user script to a Toolforge tool.

Background: I'm making a tool for watchlisting a section (a user can subscribe/unsubscribe to a section). I don't want to store subscription lists onwiki because sections can move, be merged, etc, which would require a bot to update them. Also, just as watchlists aren't public, section watchlists shouldn't be, either. Thus, I want the userscript to make authenticated requests to a tool that stores the subscription lists in a database.

I don't know how to verify (in the tool's code) that incoming requests were actually made by the user, though. OAuth won't help here, I think. Ed6767 on enwiki suggested generating a random string, storing it in a cookie, saving a single edit with the hash of the string, and then sending the string up with every request (the server would then check it). Seems fine to me, although I feel like there are other existing ways of doing this. Thoughts, please? Enterprisey (talk) 01:30, 4 July 2020 (UTC)

@Enterprisey: Can you remind me why OAuth doesn't work? --AntiCompositeNumber (talk) 02:30, 4 July 2020 (UTC)
AntiCompositeNumber: I want to verify, when the tool receives a request, that the request was made from the userscript running in a logged-in wiki session by the appropriate user. My original thought was to perform the authorization code flow for an access token, and then every time I get a request from the user, to perform an authenticated request to oauth2/resource/profile. Speculation follows. If the response is based exclusively on the validity of the access token, it'll succeed every time (barring expiration) and I get no indication as to whether there is a current session. If, however, the response from oauth2/resource/profile is based on the enwiki (for example) session-token cookies, then those cookies will not be present in the request from my tool to enwiki, so it'll fail every time and I'll also have no indication either way. Enterprisey (talk) 03:35, 4 July 2020 (UTC)
@Enterprisey: How about userjs- preferences and mw:API:Options? You could store your watchlist information there and not have to worry about authenticated requests to Toolforge. Of course, that wouldn't work quite so well for large watchlists. In that case, you could store a random token that you could hash and use as a key to the database. I'd have the user visit the tool, OAuth auth, generate the token, stick it in their prefs & the database, then throw away the OAuth creds. It's similar to Ed6767's suggestion, but I think it's a bit cleaner as it doesn't require anything to be visible on-wiki and doesn't require a new token on different devices. --AntiCompositeNumber (talk) 03:58, 4 July 2020 (UTC)
Excellent ideas! Thank you so much. I think that, or some variant, will be what I go with. Enterprisey (talk) 05:49, 4 July 2020 (UTC)

oauth2/resource/profile works every time if a valid access token is provided, so the access token will act as the user's proof of identity. You could store it as a cookie for the toolforge domain, but then you'd have to worry about public machines and whatnot; or you could force the user to fetch a new access token every time, but then they'd have to visit the toolforge tool's domain, so it won't really work with an AJAX request. (Maybe that can be worked around, but seems more effort than worth it.)

I'd probably make the tool an OAuth app with user options access, have a setup phase where the user visits the Toolforge tool's website, does the OAuth authorization, the tool verifies their identity, and generates a random password token and stores it in a userjs- option. The user script could then read the password from there and use it to authenticate requests. (Please don't put the entire watchlist in userjs. All userjs data is included in the page HTML on every pageview so that scales poorly.) It's not super elegant, someone from the WMF Security team might be able to suggest a better option. --Tgr (talk) 20:28, 4 July 2020 (UTC)

On second thought I think there would be a simpler approach: register a non-confidential OAuth 2 consumer (ie. one that can be used by a client-side app), authorize it from the gadget, then put the access token (maybe some PKI encrypted version of the access token if you want to be paranoid) into every request sent to the Toolforge tool. The tool then uses the access token to get the name of the user. This might need development on the OAuth server side though, as I think right now OAuth 2 does not allow getting an access token without going through the authorization dialog. Let's get back to that in the future. --Tgr (talk) 23:41, 12 July 2020 (UTC)

Improvement in Home Page

Can we highlight the: (pictured) word in the text which denotes the the text is related to the photo ... like, perhaps, making it bold or red?

I'm always finding myself searching for the word: (pictured) to find the text associated with the photo.

  -Thanks so much, Starmanron
We don't have any pictures on the home page. What project do you mean? Stryn (talk) 16:01, 21 July 2020 (UTC)

Interwiki mess

Hello. How to revert (newely installed?) interwiki mess? Earlier, all interwiki links were listed by prefix alphabetically so one could expect where to find link to the necesary wikiproject. Now it is mess without any logic (can not figure how they are sorted - not alphabet, not size, not region, it seems just random). Hugo.arg (talk) 17:40, 21 July 2020 (UTC)

phabricator:T257625. --MF-W 17:43, 21 July 2020 (UTC)

Weird increase in pageviews about Wikimedia projects

I wonder what happened to pageviews on articles about Wikimedia projects, see these numbers. There is a huge increase from 17 July, and it's on every language projects. Did people for example suddenly hear about Wikidata (on Finnish Wikipedia)? Stryn (talk) 15:23, 22 July 2020 (UTC)

You are right. Pageviews on Wikipedia increased by three orders of magnitude, for instance for w:en:Wikibooks, but not for w:en:Wikipedia. The effect is even greater on the main pages of the sister projects, for instance b:en:Main_Page and equivalents (now 600k views/day instead of the usual 30k/d). There is no such growth for other popular pages on the same domain.
I'd say it's clearly some bug, because it's very unlikely that Wikibooks gets half the interest of Wikipedia (itself the most popular page on the English Wikipedia), and only on the desktop site but not the mobile site. The last time something like this happened, it was due to a Safari bug, IIRC. It might also just be some undeclared automatic process though. You can file a bug report to the analytics folks to see if they can check the logs and find out more. Nemo 19:42, 22 July 2020 (UTC)

Font Size

Dear Wikipedia,

Who's idea was it to change the standard font size to 0.75em? Someone at your department seems to hate accessibility. Please revert the font size back to 1em (or rem).

Kind Regards,

Jens

Which wiki do have in mind? Ruslik (talk) 19:14, 7 August 2020 (UTC)

Switching off the visual editor globally

How do I tell the software that I don't want to use the visual editor, on any SUL wiki? Right now when editing one for the first time (or in some time, I guess), I always have to wait for that selector to load and then to choose the source editor. Or, when the wiki has a "visual edit" and a "source edit" tab, clicking away the advert for the visual editor. --MF-W 21:58, 7 August 2020 (UTC)

@MF-Warburg: Changing the editing mode to "Always give me the source editor" in w:Special:GlobalPreferences#mw-prefsection-editing (not on Meta unless you've got the beta feature on) should do it. I don't know if anything prevents the ask on first use though. --AntiCompositeNumber (talk) 22:03, 7 August 2020 (UTC)
Actually, I think the "Temporarily disable the Visual Editor while it is in beta" switch might. --AntiCompositeNumber (talk)
Thanks, I think it worked. Though for the second case I described, there is still a pop-up, but it now has only one option. --MF-W 22:14, 7 August 2020 (UTC)

article image upscaling

Hi. My name is Artyom. I’m a reader of wikipedia. I got the problem that I can’t upload my upscaled images to the article. Example: I see that some images in article is low quality. By using AI program made for lossless resolution increase. And I can’t upload this upgraded image to the article because of copyright claim. I don’t realize how that rule with copyright works so the original image was uploaded by the same person as me or others.

I suggest you to add option that allows anyone to upload improved version of special image. Something like button «upload better quality version» . And there will not be any copyright problems so responsible editors can check such edits.

Best regards,

Maltsev Artyom, Omsk Russia — Preceding unsigned comment added by 46.50.203.11 (talk) 07:33, 27 June 2020 (UTC)

CSS load cost on gomwiktionary main page

I recently created a new main page for the Konkani Wiktionary. The main page is made up of three templates, and the CSS for all three templates is applied using TemplateStyles from a single CSS page. In this case, is the CSS load cost three times the size of the common CSS page (is the CSS page being loaded once for each template), and should I split the single CSS page into three pages that contain the styles specific to each template? The Discoverer (talk) 16:51, 13 August 2020 (UTC)

@The Discoverer: The styleseet content itself is emitted only once per page, all other <templatestyle> tags are represented with <link rel="mw-deduplicated-inline-style" href="mw-data:TemplateStyles:...">. By the way, the whole stylesheet is not that big that you would need to worry about performance. —Tacsipacsi (talk) 17:24, 13 August 2020 (UTC)
@Tacsipacsi:, thank you for your reply. :) The Discoverer (talk) 17:45, 14 August 2020 (UTC)

Anti Spam

I attempted to ask a question but it shows anti spam. How to avoid? I mention the word "domain" and "redirect"

@Absad1: Hard to say without knowing which website, which page, and which exact and complete error message. (Also, please sign your posts.) --AKlapper (WMF) (talk) 14:04, 22 August 2020 (UTC)

Change that breaks layout

In [4], the "portal" css class is removed, but many scripts is still using it: Commons Meta enwiki

Instead the "vector-menu-portal" class should be used.

MediaWiki:InterProject.js is loaded by default in Meta (and some other wikis), but I doubt how it is useful nowadays.--GZWDer (talk) 02:10, 27 August 2020 (UTC)

For Commons, there appear to be only three uses: a user script used only by a now-globally-locked user, a script in the MediaWiki space of unknown use and purpose that doesn't appear to be used by anyone, and the aforementioned InterProject.js which also no longer appears to be in use on Commons. It appears that the "scream test" is the best solution here: turn it off and see who screams. --AntiCompositeNumber (talk) 02:27, 27 August 2020 (UTC)
As the commit notes, the core CSS is now depending on another classname vector-menu-portal, and yes people should switch to that, however the classname DOES remain present in the HTML for now, so scripts should NOT break, however they might miss some of the styling they depend on. —TheDJ (talkcontribs) 10:39, 27 August 2020 (UTC)

The “Other languages” header (shown during editing in place of the interlanguage link bar, probably by UniversalLanguageSelector) is also broken. I don’t want to bother with creating a Phabricator task now, can someone fix it? —Tacsipacsi (talk) 22:02, 27 August 2020 (UTC)

@Tacsipacsi: Could you elaborate which exact steps to perform on which website to see some broken "Other languages" header somewhere? Thanks! --AKlapper (WMF) (talk) 17:54, 31 August 2020 (UTC)
@AKlapper (WMF): Just open any page for editing that has ULS (e.g. any page on Meta while logged in, like this one you’re reading right now). The “Other languages” “In other languages” (sorry, I haven’t checked the exact English string last time) header at the bottom of the left-hand sidebar is unstyled. —Tacsipacsi (talk) 19:49, 1 September 2020 (UTC)
@Tacsipacsi: See phab:T262374 for unstyled ULS stuff (I think). --AKlapper (WMF) (talk) 14:32, 11 September 2020 (UTC)

Trouble with Visual Editor

Today I was trying to add a reference citation to the English Wikipedia article about Blanch Armwood, and when I had completed the Visual Editor form for "Cite News" and clicked "Submit", I received the following: Error contacting the Parsoid/RESTBase server (HTTP 503). Not sure if this was a one-time issue, or a more pervasive problem. - AppleBsTime (talk) 13:13, 9 September 2020 (UTC)

You'd better ask on en:Wikipedia:Village pump/technical. Ruslik (talk) 18:03, 12 September 2020 (UTC)

7-zip one of the best compressors?

Hi guys, it's 7-zip one of the best compressors to use in the ios system according to [redacted spammy link].

this is true or what do you recommend?

@Mribelina12: Hi and welcome. This page is about technology in Wikimedia, so discussion of unrelated software is off-topic here... (PS: Please sign your posts with your signature so it's clear who wrote something. PPS: Also see https://heyguys.cc/ ) Cheers, --Malyacko (talk) 16:00, 16 September 2020 (UTC)

Tools for compiling list of Wikimedia Commons users by contributions/uploads

Commons does not appear to ever have had an automatically updated page like the one on English Wikipedia and other wikipedias that compiles a list of top users by overall contributions.

Additionally, while it does tools that are supposed to compile lists of users by uploads, neither has been working for the over a year. one stopped updating on 31 Jan 2019. This one stopped updating 28 May 2019.

Would anyone be able to repair or replace the tools that compile lists of top user by upload counts, and create one that would compile a list of top users by overall contributions? SecretName101 (talk) 03:48, 2 September 2020 (UTC)

@SecretName101: See the link on https://stats.wikimedia.org/wikispecial/EN/TablesWikipediaCOMMONS.htm which leads to https://stats.wikimedia.org/#/commons.wikimedia.org/contributing/top-editors/normal%7Ctable%7Clast-month%7C~total%7Cmonthly --AKlapper (WMF) (talk) 16:56, 3 September 2020 (UTC)

@AKlapper: But that only provides you info as to who had the most edits on a given day or month, as far as I can tell. That doesn't tell you who has had the most edits of all-time. SecretName101 (talk) 08:28, 15 September 2020 (UTC)

@SecretName101: See "Report a bug" at the bottom of the page that I linked, and follow mw:How to report a bug. (PS: Your ping did not work as you dropped half of my username. :) --AKlapper (WMF) (talk) 13:49, 15 September 2020 (UTC)
My bad on that ping fail SecretName101 (talk) 19:31, 15 September 2020 (UTC)
@AKlapper (WMF): is it really a bug though? Because the tool is functioning to do what it was designed to. It just doesn't sufficiently supplant the former tools' use in regards to compiling an all-time list of uploaders to Commons, nor does it have a use to compile an all-time list of users by edits either. SecretName101 (talk) 19:36, 15 September 2020 (UTC)
@SecretName101: I don't see how it matters if it is a bug. --AKlapper (WMF) (talk) 10:31, 17 September 2020 (UTC)