Am Thu, Jul 16, 2026 at 12:54:34PM +0200 schrieb Hugo Buddelmeijer: > - 'outdated': if it is useful to people, it is by definition 'for today'
This is unclear already in Guix; many of those are one-off, drive-by contributions, which may not have more than the initial contributor as user. And probably it is not useful anymore to the authors, since otherwise they would work on keeping it up to date. If this work is spread over many authors, it is parallelised and manageable; if instead it is done in Guix by a handfull (rounded up) of active Python team members, it becomes bothersome. > - 'fix': these packages are not broken; they work Also not necessarily. For instance, there was the gcc-14 transition last year. Many old packages did not build with gcc-14. This was a lot of work to either update them somehow or add a number of fixes to the package recipe, or at worst compile them with an older version of gcc. This in turn requires to keep a lot of old compiler versions; leading to churn on the build farms, and also maintenance work. See, for instance, https://codeberg.org/guix/guix/issues/8882 where the struggle is to make an ever growing number of llvm versions build on the hurd. Or think of newer architectures like risc-v. I do not think that turning Guix into an eventually uncurated dump of old software is a good solution to our update problems. Andreas
