On Thu, Nov 11, 2021 at 11:16 AM Laurenz Albe <laurenz.albe@cybertec.at> wrote:
> Nobody will want to hear that, but the only really good solution would
> be for PostgreSQL to have its own built-in collations.
+1.
I agree with Tom that it sounds like a lot of work. And to be honest
it's work that I don't really feel very excited about. It would be
necessary to understand not only the bona fide sorting rules of every
human language out there, which might actually be sort of fun at least
for a while, but also to decide - probably according to some
incomprehensible standard - how Japanese katakana ought to sort in
comparison to, say, box-drawing characters, the Mongolian alphabet,
and smiley-face emojis. I think it's not particularly likely that
there are a whole lot of documents out there that include all of those
things, but the comparison algorithm has to return something, and
probably there are people who have strong feelings about what the
right answers are. That's a pretty unappealing thing to tackle, and I
am not volunteering.
On the other hand, if we don't do it, I'm suspicious that things will
never get any better. And that would be sad.
--
Robert Haas
EDB: http://www.enterprisedb.com
