| 9 Nov 2024 |
 Ramses 🇵🇸 | In reply to @acidbong:envs.net
what version and did you install it from source or from nixpkgs? I'm building from main. I update more or less daily, but this process might have been running for a while. The last time I rebooted was a week ago, so it must be running a commit from the last week on main | 08:48:52 |
| Ramses 🇵🇸 | This is on NixOS | 08:49:52 |
 aidalgol | In reply to @sigmasquadron:matrix.org
i like nixd This is really good. Thanks for the recommendation. | 19:45:58 |
 olivia | is this situation a bug?
1. I bump the flake lockfile in repo A
2. I update flake.nix in repo B to point to local repo A (with a bumped lockfile) instead of remote
3. trying to build things in repo B uses the old versions of transitive dependencies through repo A rather than the new ones
4. nix flake update B in repo A fixes it | 22:50:54 |
| olivia | (I'm not using inputs.A.follows in B) | 22:53:31 |
| olivia | is this situation a bug?
1. I bump the flake lockfile in repo A
2. I update flake.nix in repo B to point to local repo A (with a bumped lockfile) instead of remote
3. trying to build things in repo B uses the old versions of transitive dependencies through repo A rather than the new ones
4. nix flake update A in repo B fixes it | 22:53:38 |
| aidalgol | Are you committing after step 1? | 22:57:56 |
| olivia | yes | 22:58:19 |
| aidalgol | I think nix flake update A fixes it because updating the input in repo B's flake.nix will only bump the rev for that input, but the locked transitive dependencies will be left unchanged. | 23:00:51 |
| olivia | In reply to@aidalgol:matrix.org
I think nix flake update A fixes it because updating the input in repo B's flake.nix will only bump the rev for that input, but the locked transitive dependencies will be left unchanged. yeah, that is what appeared to happen | 23:01:09 |
| aidalgol | I'm not sure whether that's a bug or by design. Heck, I'm not sure anyone knows what's by design in flakes at this stage. :P | 23:01:15 |
| olivia | lol | 23:01:20 |
| olivia | yeah I asked because wasn't sure whether this was by design :( | 23:02:19 |
| olivia | (or, at least, part of the de facto stable external interface at this point) | 23:02:45 |
| olivia | I also have vague memories of bumping rev and having it bump the transitive deps in the local lockfile automatically in the past, but could be misremembering | 23:03:41 |
 KFears (annoying) | I think it is part of the stable interface, yeah :(
IIRC there are some issues in CppNix repo that have been up for a really long time about that, but I might be misremembering, better check by yourself | 23:04:09 |
| olivia | https://github.com/NixOS/nix/issues/8976 looks like this | 23:05:35 |
| olivia | that issue has at least not been closed yet | 23:06:47 |
| olivia | oh and a corresponding lix isssue and CL :) | 23:08:36 |
| olivia | hmm... the CL is merged though | 23:09:00 |
| KFears (annoying) | There's this:
https://github.com/NixOS/nix/pull/8977
https://github.com/NixOS/nix/issues/8976
And this:
https://github.com/NixOS/nix/pull/9173
https://github.com/NixOS/nix/issues/9143 | 23:10:33 |
| olivia | oh nvm I missed that the CL is just a workaround for lix's CI or similar | 23:10:42 |
| KFears (annoying) | I'm trying to understand what's the difference between those, hmmm | 23:10:44 |
| KFears (annoying) | Ah, okay, the first seems to be the issue you're hitting, and the second is some weird special case with inputs.foo.inputs.bar.url | 23:15:27 |
| olivia | that's my understanding as well | 23:16:13 |
| KFears (annoying) | I'll see if I'm able to fix this | 23:16:16 |
| KFears (annoying) | I'm a little hesitant if that should be fixed, because that's part of the flake quasi-stable interface, but it should be fine? The part that experiences issues is CLI, and it's thankfully a lot less design-tied than other things | 23:17:39 |
| KFears (annoying) | And not updating transitive inputs is probably a lot more broken than doing it :/ | 23:18:08 |
| olivia | I cannot imagine a situation where the current behavior would be desirable | 23:18:21 |
| aidalgol | It's certainly very confusing. | 23:18:28 |
