On 11/11/2021 08:14, Michael Paquier wrote:
> I would suggest to send any improvements for imath directly to
> upstream, then Postgres would just feed on that when we update it:
> https://github.com/creachadair/imath
I didn't realise imath comes (occasionally) from external repo. Thank you.
The compilation warnings are only relevant for their old releases,
before imath v1.27 where they've replaced macros with inline functions, see
https://github.com/creachadair/imath/commit/d4760eef8
So imath in upstream doesn't require updates (especially since those
warnings come specifically from pgindent's work, not from the upstream
source). But Postgres' 10_STABLE and 11_STABLE still use old version
with CLAMP as a macro.
Noah Misch was last to update imath from upstream in February 2019 and
that change became part of REL_12_STABLE and later:
> Import changes from IMath versions (1.3, 1.29].
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=48e24ba6b7fd3bfd156b51e8d768fd48df0d288b
Noah and others, do you think it would be a good idea to update imath in
REL_10_STABLE and REL_11_STABLE to a later upstream version to avoid
those warnings about misleading indent? (see beginning of this thread)
Admittedly, those warnings can only be seen in unusual circumstances (no
openssl), but still, as you have mentioned in your commit message,
> We're better off naively tracking upstream than reactively
> maintaining a twelve-year-old snapshot of upstream.
Perhaps even update imath in all relevant stable branches, 10..14
inclusive, to the same upstream version?
On 11/11/2021 08:14, Michael Paquier wrote:
> Peter has removed this code as of db7d1a7, but there could still be a
> point in updating imath on the stable branches where it exists if
> there is an issue with it impacting pgcrypto without OpenSSL.
Perhaps that would be a good idea, see above.
--
Anton Voloshin
https://postgrespro.ru
Postgres Professional, The Russian Postgres Company