On 7/7/26 18:59, Ludovic Courtès wrote:
Hi Hugo,
Hugo Buddelmeijer via "Development of GNU Guix and the GNU System distribution."
<[email protected]> skribis:
Revisiting this GCD seems impossible to me when there are people in
the community who rather have Guix die.
Please note that I was quoting what others explicitly said earlier in
the discussion. I would personally never threaten to let Guix die over
such a disagreement.
I think what we all have in common, those of us who participated in the
discussion, is that we care about Guix.
Yes, I do assume good faith. I even consider it a valid position to
prefer death over compromising values. I copied the quote because such
an extreme position does make it very hard to build consensus to change
the GCD in the future.
That is, I reacted to your statement that we can always amend the GCD
because I found that disingenuous, since you know these extreme
viewpoints exist; it was not a reflection on the viewpoint itself.
(And I don't expect a better "guix refresh" to be feasible; I'd be
happily proven wrong though, and will do my part to get there.)
Respectfully, I think you’ve been proven wrong by Guix-CRAN, by Nixpkgs,
etc. We have to be humble and learn from these experiences.
CRAN is not representative. Less than 9% of the packages in cran.scm
have custom phases or patches. It is presumptuous (as antonym to
humble) to assume that approach works for other ecosystems; more than
25% of all packages have custom phases or patches.
I'd be happily proven wrong though, and am willing to work on improving
the situation.
having contributed to ‘guix refresh’ and to some of the
importers/updaters, I find this kind of work pretty fun and challenging.
It can also be fun and challenging to fix broken packages, and there is
plenty of fun to go around.
E.g., anyone who'd like to take a shot at refreshing onnx? Onnx is
currently blocking upgrading NumPy (since onnx uses numpy and is itself
used by calibre, which I consider important):
https://codeberg.org/guix/guix/issues/9443
Hugo
