Re: Built-in CTYPE provider

On Thu, 2024-07-18 at 13:03 -0400, Tom Lane wrote:
> In the second place, I cannot understand why pg_c_utf8 is being
> held to a mutability standard that we have never applied to any
> other locale-related functionality --- and indeed could not do
> so, since in most cases that functionality has been buried in
> libraries we don't control.

I believe that we should hold it to a higher standard *precisely
because* the previous way that we handled mutability in locale-related
functionality was a problem.

> It seems to me to be already a
> great step forward that with pg_c_utf8, at least we can guarantee
> that the behavior won't change without us knowing about it.

+1

But the greatness of the step depends on our readiness to be careful
with such changes.

> Noah's desire to revert the feature makes the mutability situation
> strictly worse, because people will have to continue to rely on
> OS-provided functionality that can change at any time.

I think everybody agrees that we don't want to expose users to data
corruption after an upgrade.

It understand Noah to take the position that anything less than
strict immutability would be worse than the current state, because
currently a packager can choose to keep shipping the same old
version of libicu and avoid the problem completely.

I don't buy that. First, the only binary distribution I have heard
of that does that is EDB's Windows installer. Both the RPM and
Debian packages don't.

And until PostgreSQL defaults to using ICU, most people will use
C library collations, and a packager cannot choose not to upgrade
the C library.

I believe the built-in CTYPE provider is a good thing and a step
forward. But to make it a big step forward, we should be extremely
careful with any changes in major releases that might require
rebuilding indexes.
This is where I side with Noah.

Yours,
Laurenz Albe