Re: ICU locale validation / canonicalization

On 13.03.23 16:31, Jeff Davis wrote:
>> What we had discussed a while ago in one of these threads is that ICU
>> before version 54 do not support keyword lists, and we have custom
>> code
>> to do that parsing ourselves, but we don't have code to do the same
>> for
>> language tags.  Therefore, if I understand this right, if we
>> automatically convert ICU locale IDs to language tags, as shown
>> above,
>> then we break support for such locales in those older ICU versions.
>
> Right. In versions 53 and earlier, and during pg_upgrade, we would just
> preserve the locale string as entered.

Another issue that came to mind: Right now, you can, say, develop SQL
schemas on a newer ICU version, say, your laptop, and then deploy them
on a server running an older ICU version. If we have a cutoff beyond
which we convert ICU locale IDs to language tags, then this won't work
anymore for certain combinations. And RHEL/CentOS 7 is still pretty
popular.