Re: ICU fixed digit reordering bug

On Tue, Oct 27, 2026 at 2:40 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> Peter Eisentraut <peter(at)eisentraut(dot)org> writes:
> > There have been various reports over time that ICU collations that
> > reorder digits are buggy. ...
> > This issue was finally fixed in ICU 79, which came out a few days ago.
> > I have confirmed that the PostgreSQL issue is fixed with that new version.
>
> > The question now is whether there is anything else we want to do about this?
>
> > 0. Do nothing. Leave the ICU bug and possibly this email as reference
> > and recommend that users don't use the affected kinds of ICU locales
> > with the affected older ICU versions.
>
> I like option 0. It can't be incumbent on us to provide defenses
> for every bug in every piece of software that we depend on ...
> especially not edge-case bugs like this one seems to be.
>
> If we had a surplus of time and people-power available, maybe it
> would be reasonable to do something about this. But you know as
> well as I do how horribly short we are on both.
>
> regards, tom lane
>
>
One upgrade concern may be worth calling out: if a PostgreSQL installation
moves to ICU 79, comparisons for a digit-reordering collation will use the
corrected behavior, while an existing index may have been built under the
old behavior. That could leave the index inconsistent with current comparisons
and lead to wrong query results until it is rebuilt.

Should we explicitly advise users of affected collations to REINDEX after
upgrading ICU? It would also be useful to confirm whether the collation
version check reliably warns for this particular fix.

Regards,
Haibo