3.4.3

[Greg Troxel]

"Regina Obe" <lr at pcorp.us> writes:

> Why are we wasting our time with rc1, we never do rcs for a micro
> release.  Just release will ya 😊

My experience is that proving lack of regression on platforms you didn't
test on is basically impossible.  It helps if there is a formal rule
that changes on micro branches are limited to bug fixes, and those are
checked for any newly-used functions being required by documented
specifications.

Packagers basically test things that are structurally like releases.
That verifies the "get tarball, unpack, build" process, on something
that is really like a release, that was created via the same pathway.

If there were RCs and nobody tested, repeatedly, then it would be silly.
But I try to update pkgsrc (locally) to each alpha/beta/rc.  Generally,
that makes me do the update work sooner, but the total work isn't really
that much more.  (e.g. the geos beta2 was just change the number, 'make
distinfo', 'make replace', 'mkae test').

> If there are issues you will push out a 3.4.4

I have had experiences where it works on linux and not on some less
common platform and people say that there is no need for a new micro
because.. it works on linux.  So maybe in postgis, but in general this
claim of a plan is not credible.

The real issue though, is that doing things in a rush is bad, and
emergencies are bad.  I don't think 24h or 72h after pg17 is a big deal.

But I'm already testing, including updating to pgsql16 on the way...