Previous discussion: https://meta.wikimedia.org/wiki/Talk:Interwiki_map/Archives/2020#Interwiki_prefixes_that_conflict_with_language_codes

aew and twl are also prefixes matching existing language tags.

The other 3-letter prefixes that haven't been assigned as language codes yet are dpd, hdl, irc, rev, rfc, svn, wmf and wqy.

By "valid format", do you mean that it's well-formed (it has the right format to potentially be a language tag, even if it (or part of it) is not assigned right now)?

Over half of the current interwiki prefixes would be well-formed language tags - the language code part of a language tag can be 2 to 8 letters long (see https://www.rfc-editor.org/rfc/rfc5646.html#section-2.1).

In practice, it's unlikely that there will be any new 2-letter codes, 4-letter codes are only reserved for potential future standards, and they're unlikely to approve any 5-8 letter codes either (search the ietf-languages mailing list archives for Elfdalian if you're interested), which only leaves 3-letter codes.

I think it would make more sense to avoid prefixes which are 3 letters long - that's much simpler to implement (doesn't need a copy of the language subtag registry), would cover the existing cases, and would limit it to codes which might be used as interwiki language prefixes (Langcom are unlikely to approve any wikis using language codes with subtags).

Note that the task title here conflates https://meta.wikimedia.org/wiki/Interwiki_map (interwikis used on Wikimedia wikis) with https://gerrit.wikimedia.org/g/mediawiki/core/%2B/HEAD/maintenance/interwiki.list (default interwikis for new MediaWiki wikis). Updates to the former are managed on-wiki and not via Phabricator. Updates to the latter are managed via Phabricator.

On specific issues:

• dcc is going to be deleted soon anyway for unrelated reasons (https://meta.wikimedia.org/wiki/Talk:Interwiki_map#Dcc)
• doi is realisitically never going to be updated or removed. In the unlikely event a Dogri Wikipedia gets created we will have a T363538 again.
• Filed git as https://meta.wikimedia.org/wiki/Talk:Interwiki_map#git
• Filed zum as https://meta.wikimedia.org/wiki/Talk:Interwiki_map#zum
  • Personally as a Meta admin I am not willing to implement either of these myself. Others may feel differently.

Filed aew and twl over on Meta as well.

In T360792#9751638, @Pppery wrote:

Note that the task title here conflates https://meta.wikimedia.org/wiki/Interwiki_map (interwikis used on Wikimedia wikis) with https://gerrit.wikimedia.org/g/mediawiki/core/%2B/HEAD/maintenance/interwiki.list (default interwikis for new MediaWiki wikis). Updates to the former are managed on-wiki and not via Phabricator. Updates to the latter are managed via Phabricator.

On specific issues:

• dcc is going to be deleted soon anyway for unrelated reasons (https://meta.wikimedia.org/wiki/Talk:Interwiki_map#Dcc)
• doi is realisitically never going to be updated or removed. In the unlikely event a Dogri Wikipedia gets created we will have a T363538 again.
• Filed git as https://meta.wikimedia.org/wiki/Talk:Interwiki_map#git
• Filed zum as https://meta.wikimedia.org/wiki/Talk:Interwiki_map#zum
  • Personally as a Meta admin I am not willing to implement either of these myself. Others may feel differently.

Citing 2020 discussion in Meta for reference (by PiRSquared17): Dogri has an open request for a Wikipedia. However, doi is a macrolanguage code. The individual codes are dgo (Dogri proper), xnr (Kangri). Crisis averted, probably.