Adding Pablo as the IM in charge for the ongoing incident, Lydia as the PM for Wikidata, Thiemo as the dev that provided the patch and Adam as the one in charge of tech tickets.

I believe this caused both T259744 and T259745. I already created T259779 where I describe an idea that is a little different from "should always and in en". One of the two tickets might be a duplicate now.

This is not technically a Regression as it was always broken for "languages" like ⧼Lang⧽ or %E2%A7%BCLang%E2%A7%BD . However, since we have validation for our TermLanguageFallbackChain, this problem/bug has become more apparent, because now the chain can be empty whereas before it at least contained the invalid interface language.

Also, I think this is still causes rare-ish production errors. While they may not have a big impact, they do show up in our logs. Confusingly, they only show up since the last train on August 12th, not since the week before as I would have expected: T260384: Wikimedia\Rdbms\Database::makeList: empty input for field wbxl_language

Change 620689 had a related patch set uploaded (by Lucas Werkmeister (WMDE); owner: Lucas Werkmeister (WMDE)):
[mediawiki/extensions/Wikibase@master] Inject MediaWiki language services into LanguageFallbackChainFactory

https://gerrit.wikimedia.org/r/620689

Change 620689 merged by jenkins-bot:
[mediawiki/extensions/Wikibase@master] Inject MediaWiki language services into LanguageFallbackChainFactory

https://gerrit.wikimedia.org/r/620689

Is this done?

In T259783#6390525, @Ladsgroup wrote:

Is this done?

Unfortunately: no. This was only a minor refactoring removing some calls to deprecated methods. The underlying problem that this ticket is about (and that might be the source of some production errors/logspam) is still undone and actually quite tricky. But we are on it :)

Picked up in storytime, with the plan of initially creating an investigation ticket.

Change 621262 had a related patch set uploaded (by Michael Große; owner: Michael Große):
[mediawiki/extensions/Wikibase@master] Fix Term Fallback chains being empty

https://gerrit.wikimedia.org/r/621262

Change 621262 merged by jenkins-bot:
[mediawiki/extensions/Wikibase@master] Fix Term Fallback chains being empty

https://gerrit.wikimedia.org/r/621262

moving back to doing for the testcase

Change 621305 had a related patch set uploaded (by Michael Große; owner: Michael Große):
[mediawiki/extensions/Wikibase@master] DNM: Account for empty fallback chain on Unicode language string

https://gerrit.wikimedia.org/r/621305

I'm somewhat stuck with adding the test in this patch: https://gerrit.wikimedia.org/r/c/mediawiki/extensions/Wikibase/+/621305
The trouble is that I can't think of a way to actually test that the mock is adjusted that wouldn't also be green if it weren't.

Change 621305 merged by jenkins-bot:
[mediawiki/extensions/Wikibase@master] Account for empty fallback chain on Unicode language string

https://gerrit.wikimedia.org/r/621305

When this is deployed with the next train, then we should check whether T260384 still occurs. (logstash)

I think this can now be closed: