Re: C99

[Jason Bacon <jtocino%gmx.com@localhost>]

On 3/7/25 05:04, Benny Siegert wrote:
> > It wouldn't help me, it would make things worse.  I'd now have a 2 - 3
> > week period in which the pressure is on to fix all of the breakage from
> > the past 3 months, before I can then start on any development work I'd
> > need to do.  After which I'd then need to go and sync back to my fork
> > and continue there anyway, as pkgsrc will be broken again for me within
> > days after the freeze finishes and the floodgates of untested changes
> > are resumed.
>
> I see.
>
> At the macro level, what would help with this is if pkgsrc had CI. CI
> that is more granular than running full bulk builds and seeing that
> things started failing. Some system that would check each commit for
> breakage, with some sort of consequences. But that would require a
> very different development model. Something like each commit being on
> a branch and only being merged after testing.

That's exactly what pkgsrc would need, and exactly what Jonathan has set
up for dreckly.  It's hard to change a culture, though.  For anyone who
wasn't aware, Jonathan made a CI system for pkgsrc available before
forking dreckly, but few people showed interest.

As my primary platforms of interest are Apple Silicon and RHEL, which
have always suffered frequent regressions in pkgsrc, I've turned my
focus to dreckly as well.  I can't realistically promote pkgsrc to
people in the scientific community, knowing that a routine update is
likely to break the tools they need for their research.

It's very unfortunate that it has come to this, but sometimes a fresh
start is the only real solution.

--
Life is a game.  Play hard.  Play fair.  Have fun.