On 7/16/26 12:17, Andreas Enge wrote:
Am Thu, Jul 16, 2026 at 12:06:23PM +0200 schrieb Hugo Buddelmeijer via
Development of GNU Guix and the GNU System distribution.:
Most of that time was spend on fixing otherwise
unrelated packages to work with Python 3.12. Like one-off old
bioinformatics packages.
Caeterum censeo that we should more aggressively remove outdated
packages (especially one-offs that their authors do not bother to fix);
A week ago, I would have said the same thing. But I've come around and
now believe this exactly the opposite of what we should do. (But for
now, I'm fine with removing more packages.)
Related, I (now) object to the word choice:
- 'fix': these packages are not broken; they work
- 'aggressively': this is akin to enforcing innovation
- 'bother': the forced churn is what is bothersome, not their reluctance
- 'outdated': if it is useful to people, it is by definition 'for today'
We can keep those older packages around without 'fixing' them, and also
without them interfering too much with newer packages.
if need be, they can move to guix-past, or be used via the time machine.
The deprecation policy allows for this.
I think it would be great if we can make `guix time-machine` and
inferiors even easier to use. That is a big strength of Guix.
I went all-in on Guix once I learned how to fix such dependency problems
autonomously through inferiors. In retrospect, that was the point of no
return for me.
Maybe we even want to have in the main Guix repository an easy mechanism
to say "this package uses dependencies A, B, C from commit xyz".
