You're viewing the f.duriansoftware.com public feed.

d@nny disc@hipsterelectron@circumstances.run
Oct 14, 2026, 3:38 PM
@miss_rodent @Lightfighter @TodePond i have worked on several systems which avoid this and unfortunately the resolver needs to become much more complex to fix it paticularly since there is essentially no formal theory of this and most modern ones just fling shit at an SMT solver and let the user figure it out when it fails which is a terrible experience. i think what spack has is the best possible thing and they care a lot about user experience but i think we can do better
💬 1🔄 0⭐ 0

Replies

Vmiss_rodent@girlcock.club
Oct 14, 2026, 3:48 PM
@hipsterelectron @Lightfighter @TodePond Fair, don't mean to frame it as an easy or solved problem; it seems less of an issue nowadays on the user end for the big distro package managers (will defer to your judgement of how good current solutions are, it's a bit out of my scope), but, it was for a while a selling point of slackware's approach of 'You could just not." lol
💬 1🔄 0⭐ 0
d@nny disc@hipsterelectron@circumstances.run
Oct 14, 2026, 3:50 PM
@miss_rodent @Lightfighter @TodePond noooooo i think it TOTALLY validates your position actually. i'm saying this as someome who loves graph theory and realizes the problem we want it to solve is very hard to parameterize
💬 1🔄 0⭐ 0
Vmiss_rodent@girlcock.club
Oct 14, 2026, 4:00 PM
@hipsterelectron @Lightfighter @TodePond Fair, yeah it seems like a difficult thing to manage - especially if you don't control the packages themselves (and thus, can't do things like, ensure no package specifies a *specific* version of a library, rather than "versions X or later", or that no packages depend on features/behaviours that were removed after some version, etc.)
💬 0🔄 0⭐ 0