Re: Built-in CTYPE provider

On 01.04.24 21:52, Jeff Davis wrote:
> On Tue, 2024-03-26 at 08:04 +0100, Peter Eisentraut wrote:
>> The patch set v27 is ok with me, modulo (a) discussion about initcap
>> semantics, and (b) what collation to assign to ucs_basic, which can
>> be
>> revisited later.
>
> Attached v28.
>
> The remaining patches are for full case mapping and PG_UNICODE_FAST.
>
> I am fine waiting until July to get these remaining patches committed.
> That would give us time to sort out details like:
>
> * Get consensus that it's OK to change UCS_BASIC.
> * Figure out if we need a pg-specific locale and whether
> PG_UNICODE_FAST is the right name.
> * Make sure that full case mapping interacts with regexes in a sane way
> (probably it needs to just fall back to simple case mapping, but
> perhaps that's worth a discussion).
> * Implement case folding.
> * Implement a more unicode-friendly TITLECASE() function, which could
> offer a number of options that don't fit well with INITCAP().
> * Figure out if UPPER()/LOWER() should also have some of those options.
>
> Thoughts?

Yeah, I think it's good to give some more time to work out these things.
The features committed for PG17 so far are solid, so it's a good point
to pause.