9 Oct 2024 |
jade_ | In reply to @piegames:flausch.social can you link to it again please? lost the link https://gerrit.lix.systems/c/lix/+/2035 submimtted | 20:37:36 |
KFears (tragedy arc) | In reply to @jade_:matrix.org does this mean we will break it? no. we can't break it. we also can't change it. we can't improve it. So what can be done moving forward? I imagine that flakes as they are now should be, like, incompatibly rewritten into something sane, but the codebase is already very conditional on flakes-enabled vs flakes-disabled | 20:58:26 |
KFears (tragedy arc) | Is it possible to throw away flakes-disabled to just have flakes-enabled as this horrible monstrosity that you try to isolate from everything else as much as possible? Or how would that work? | 20:59:08 |
jade_ | i don't know. we can continue ignoring the issue and working on things that actually matter and leave flakes alone, which will work for at least 6 months | 20:59:12 |
jade_ | In reply to @kfears:matrix.org Is it possible to throw away flakes-disabled to just have flakes-enabled as this horrible monstrosity that you try to isolate from everything else as much as possible? Or how would that work? NO | 20:59:16 |
jade_ | flakes are not real and cannot hurt you | 20:59:27 |
KFears (tragedy arc) | So, the other way around? | 20:59:35 |
jade_ | https://jade.fyi/blog/flakes-arent-real/ | 20:59:37 |
jade_ | there is basically nothing actually conditional on flakes enabled | 20:59:48 |
jade_ | there is "the new cli does not let you use it without nix-command enabled". but like. that's not the thing that causes flake maintenance pain at all | 21:00:13 |
KFears (tragedy arc) | Wait, so why have them at all? | 21:00:14 |
jade_ | because half of the ecosystem including us depend on them | 21:00:29 |
KFears (tragedy arc) | Like, can't you just like say "use flake-compat" and just not do flakes? | 21:00:35 |
jade_ | no, flake-compat is a joke | 21:00:42 |
KFears (tragedy arc) | Figured | 21:00:48 |
jade_ | it's missing a bunch of obscure features that will cause bug reports if they are missing. imo the status quo is fine except insofar as we cannot touch the flakes code safely | 21:01:21 |
jade_ | which does suck, but there is so much else to do | 21:01:46 |
jade_ | flakes, in spite of being extremely broken, are merely an entry point into the other 90% of lix that is also very busted and in ways that actually matter to large scale usage and overall maintainability much more | 21:02:27 |
jade_ | we can leave flakes in amber and still have plenty of useful things to do | 21:02:35 |
KFears (tragedy arc) | Hopefully some path forward will appear once "everything else" is done | 21:03:06 |
KFears (tragedy arc) | My condolences for having to work on all of this | 21:03:24 |
KFears (tragedy arc) | It sounds like the legacy codebase from hell, straight up | 21:03:45 |
jade_ | i also hate that in spite of being an absolute mess, flakes are a moving target in cppnix and continue to gain more features while simultaneously being unmaintainable. | 21:08:21 |
jade_ | https://git.lix.systems/lix-project/lix/src/370ac940dd7816ad4052fafa4e0f8d17784fa16b/src/libfetchers/fetchers.cc#L135-L148 type of compat code | 21:08:45 |
KFears (tragedy arc) | In reply to @jade_:matrix.org https://git.lix.systems/lix-project/lix/src/370ac940dd7816ad4052fafa4e0f8d17784fa16b/src/libfetchers/fetchers.cc#L135-L148 type of compat code That doesn't seem too bad | 21:21:31 |
jade_ | no, it is ridiculous actually | 21:21:43 |
jade_ | it is that a test is relying on our stderr output | 21:21:48 |