[Intl] Language code 2B/2T

[SpiGAndromeda]

According to ISO 639-2 there are two sets of 3-letter language codes: 2B and 2T.

They only differ in a couple of entries. The documentation is missing information which of these sets is used (and I couldn't figure it out in the code either).

[javiereguiluz]

Maybe @ro0NL, who knows a lot about this Intl/ICU data, can help us here. Thanks!

[ro0NL]

@SpiGAndromeda AFAIK there are only 2T entries in symfony/intl.

[ro0NL]

@javiereguiluz see also https://github.com/symfony/symfony/blob/61ecf9bf6a934c5df680008daa7b6642742f8bf2/src/Symfony/Component/Intl/Data/Generator/LanguageDataGenerator.php#L188-L202

we only check "overlong", thus skiping "bibliographic" (legacy)

https://github.com/unicode-org/icu/blob/bbd72fb93e03eb3d059a8ac48af8eb25dd866ae2/icu4c/source/data/misc/metadata.txt#L383-L386

i assume all 3-long language codes in https://github.com/unicode-org/icu/blob/b9311e2f547251807c3982e6cb003feec3ba5c83/icu4c/source/data/lang/en.txt#L43 are 2T, unless we find a false positive :)

[ro0NL]

one edge case im aware of is ase - https://github.com/unicode-org/icu/blob/b9311e2f547251807c3982e6cb003feec3ba5c83/icu4c/source/data/lang/en.txt#L79

which qualifies ISO 639-3

[javiereguiluz]

Roland, thanks for the detailed info. So, can we say in the docs that we use 2T ?

[ro0NL]

IMO yes

[SpiGAndromeda]

@ro0NL I made two Pull Requests to specify the language codes (unfortunatly the Language Constraint was intruduced in 5.1 and not 4.4)

Closed the PR for 5.1 and created one for 5.4

[ro0NL]

@SpiGAndromeda very nice, this means any false positive is considered a bug now. Eg. ase, we can exclude them in symfony/symfony if you find this relevant :)

[javiereguiluz]

Fixed by #15915 and #15913. Thanks!