Re: CREATE COLLATION does not sanitize ICU's BCP 47 language tags. Should it?

On Mon, Sep 25, 2017 at 09:36:44AM -0700, Peter Geoghegan wrote:
> On Mon, Sep 25, 2017 at 9:06 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> >> The big concern I have here is that this feels a lot like something that
> >> we'll regret at leisure, if it's not right in the first release. I'd
> >> much rather be restrictive in v10 and then loosen the rules later, than
> >> be lax in v10 and then have to argue about whether to break backwards
> >> compatibility in order to gain saner behavior.
> >
> > I think it's inevitable that a certain number of users are going to
> > have to cope with ICU version changes breaking stuff.
>
> Wasn't the main point of adopting ICU that that doesn't happen when it
> isn't essential? There will be some risk with pg_dump across ICU
> versions, due rare to political changes. But, on pg_upgrade, the old
> collations will continue to work, even if they would not have appeared
> at initdb time on that ICU version, because ICU has all kinds of
> fallbacks.

I wouldn't describe it that way. I agree that few, if any, ICU upgrades will
remove country or language codes. Overall, though, almost every ICU upgrade
will be difficult. Each ICU release, even a minor release like 58.2, changes
the sorting rules in some tiny way. You then see "Rebuild all objects
affected by this collation" messages.