On Wed, Apr 19, 2023 at 1:30 PM Jeff Davis <pgsql@j-davis.com> wrote:
> On Wed, 2023-04-19 at 07:48 +1200, Thomas Munro wrote:
> > Many OSes have a locale with this name. I don't know this history,
> > who did it first etc, but now I am wondering if they all took the
> > "obvious" interpretation, that it should be code-point based,
> > extrapolating from "C" (really memcmp order):
>
> memcmp() is not the same as code-point order in all encodings, right?
Right. I wasn't trying to suggest that *we* should assume that, I was
just thinking out loud about how a libc implementor would surely think
that a "C.encoding" should work, in the spirit of "C", given that the
standard doesn't tell us IIUC. It looks like for technical reasons
inside glibc, that couldn't be done before 2.35:
https://sourceware.org/bugzilla/show_bug.cgi?id=17318
That strengthens my opinion that C.UTF-8 (the real C.UTF-8 supplied by
the glibc project) isn't supposed to be versioned, but it's extremely
unfortunate that a bunch of OSes (Debian and maybe more) have been
sorting text in some other order under that name for years.
> I've been thinking that we should have a "provider=none" for the
> special cases that use memcmp(). It's not using libc as a collation
> provider; it's really postgres in control of the semantics.
Yeah, interesting idea.