On 2/15/22 13:47, Robert Haas wrote:
On Tue, Feb 15, 2022 at 10:39 AM Joe Conway <m...@joeconway.com> wrote:
(moving to hackers)
I guess shame on me for not noticing the thread, but I don't see any
discussion about the potential for breakage to external projects.
scanint8() is exported, and this change breaks at least two extensions I
maintain.
A quick scan (no pun intended ;-)) of github shows other potential
breakage, including at least older (still open source) versions of
pglogical.
I don't have a particularly strong view on whether the underlying
change was a good idea here, but the breakage you're talking about
seems pretty easy to fix, unless I'm missing something? I think it
would be a bad idea to make such a change in a minor release without
concern for extension breakage, but in a new major release it doesn't
seem like a problem to me.
I'm not saying it is hard to fix, but it seems a bit cavalier to not
even discuss the potential for breakage. If nothing else we should flag
this as something for the release notes.
Joe
--
Crunchy Data - http://crunchydata.com
PostgreSQL Support for Secure Enterprises
Consulting, Training, & Open Source Development
