Page MenuHomePhabricator

Disable cyrillic-latin-arab converter in Kazakh Wikipedia
Closed, ResolvedPublic

Event Timeline

Ladsgroup renamed this task from Remove cyrillic-latin-arab converter in Kazakh Wikipedia to Disable cyrillic-latin-arab converter in Kazakh Wikipedia.Nov 7 2023, 1:35 PM
Ladsgroup updated the task description. (Show Details)
Ladsgroup updated the task description. (Show Details)
Ladsgroup subscribed.
Winston_Sung subscribed.
In the task description of T350684, @mukhamejan wrote:

Please note some of the bugs in the screenshot no longer existed (already fixed)


Fixed:


image.png (455×467 px, 37 KB)

Based on the current language fallback chain, it's likely not a good option to drop kk-latn from user language option (a dependency of kaa ).

Also,

We could still consider to drop regional variants as kk actually have no regional term conversions ( kk-cn , kk-kz , kk-tr ).


T39617: Do not convert text marked as being in another language with a lang attribute


("expected behavior" needed)


This is because currently the language variant code kk in KkConverter represent "mix-script" instead of only kk-cyrl .

We could probably

  • Change how the converter works
  • Use kk instead of kk-cyrl on translatewiki.net (require migration)

(tagging Language-Team about dropping language codes from user preferences, decide whether to migrate translations)

@Winston_Sung thanks for your replies. Even if these bugs are fixed, most importantly, the alphabet used in the converter is not up-to-date, no one uses it or understands it. Kazakhs in China write a completely different Arabic script, and in Kazakhs in Turkey and Uzbekistan etc, a different Latin alphabet. Therefore, the converter used in Wikipedia is not relevant. and more harm than good. The translated messages into kk-arab (54%) and kk-latin (47%) were carried out by some FuzzyBot bot, and an incomprehensible alphabet was also used.

For me, it's better to make it usable rather than dropping it.

Kazakh language in China write a completely different Arabic script
Kazakh in Turkey and Uzbekistan etc., a different Latin alphabet

So there are regional terms / different transcriptions in Kazakh language?

incomprehensible alphabet was also used

Does that mean there are old & new transliteration solutions/rules?

incomprehensible alphabet was also used

Does that mean there are old & new transliteration solutions/rules?

Yes, at the moment there are several versions of the Latin alphabet in Kazakhstan, but none of them have been approved. All three options, which are used in some places, are completely different from those used in the Wikipedia converter.

For me, it's better to make it usable rather than dropping it.

Are we planning to have another Chinese language convertor mess (T59106#8926248) here?

If a community explicitly doesn't want a broken feature, we can't force them to use it. Please stop.

Are we planning to have another Cantonese language convertor mess here?

Of course not.

T59106#8926248

That was caused by the bad conversion table and interface message caching issues.

I'm okay with completely dropping it, but it seems not that easy.

Change 972472 had a related patch set uploaded (by Ladsgroup; author: Amir Sarabadani):

[mediawiki/core@master] Drop language coverter for Kazakh

https://gerrit.wikimedia.org/r/972472

Change 972474 had a related patch set uploaded (by Ladsgroup; author: Amir Sarabadani):

[mediawiki/extensions/Wikibase@master] Drop Kazakh language variants

https://gerrit.wikimedia.org/r/972474

Change 972472 had a related patch set uploaded (by Ladsgroup; author: Amir Sarabadani):

[mediawiki/core@master] Drop language coverter for Kazakh

https://gerrit.wikimedia.org/r/972472

This patch did two thing: (1) remove the converter; (2) drop Kazakh variants as interface language. Removing a converter does not mean we need to also remove the interface language (we have de-ch and en-gb), and if we decided to remove the interface language (which I will oppose), there are more to do (language fallback chain, existing localization, translatewiki, Wikidata, etc.)

This can be turned back as an extension if needed by third party users.

What we need instead is a "plug-in" mode: MediaWiki provides a general interface for language converter and allow each converter to register itself so that we can choose whether to enable. This also allows we develop a Yue converter, only enable it in Beta Cluster, and deploy to production once it is mature.

Change 972472 had a related patch set uploaded (by Ladsgroup; author: Amir Sarabadani):

[mediawiki/core@master] Remove language coverter for Kazakh

https://gerrit.wikimedia.org/r/972472

Change 972472 merged by jenkins-bot:

[mediawiki/core@master] Remove language coverter for Kazakh

https://gerrit.wikimedia.org/r/972472

Ladsgroup claimed this task.

This will be deployed in a week or two.