Hi,
On Wed, 05 Aug 2020 16:49:49 +0200
"Daniel Verite" <daniel@manitou-mail.org> wrote:
[...]
> I'm wondering whether there is a discrepancy between the binary
> sort keys and results of direct comparisons by ucol_strcoll()
> for this collation.
[...]
This bug has been discussed back in 2018 and a customer hit it few weeks ago
as well. Here is my last analysis, various link to histories, a workaround and a
even an humble patch proposal to discuss:
https://www.postgresql.org/message-id/flat/20200610002933.6a6d482b%40firost#acba071ed158e0c03ea9a106d9b0fd6d
I even sent a patch to the ICU community to add a regression test exposing the
bug, but I had no answer.
Hopefully the workaround is good enough for you.
Regards,