9 Oct 2024 |
K900 | It was always broken | 22:53:33 |
jade_ | oh, well, problem in user code, not my problem thankfully | 22:53:59 |
K900 | https://github.com/NixOS/nix/issues/11503 | 22:54:15 |
K900 | If you want the full story | 22:54:21 |
K900 | And the short version is that we can't actually use filesets for anything a user may want to nixos-install | 22:54:52 |
K900 | And fixing it will require breaking existing eval | 22:55:28 |
jade_ | pfft | 22:55:50 |
jade_ | oh no is this reliance on toString of a path value | 22:56:34 |
jade_ | that would be both hilarious and unsurprising if so | 22:56:53 |
10 Oct 2024 |
jade_ | raitobezarius: it appears that the buildbot status reporter is not just broken in terms of usability but dangerously broken in terms of functionality: it reports Verified+1 probably on nix-eval completion before finishing building things.
also we have several instances of some kind of nixos test getting wedged in unknown ways, which might be a lix bug or something else haunted: https://buildbot.lix.systems/#/builders/1/builds/6000
really have to catch it in the act and gdb it i reckon
| 02:55:52 |
V. 🏳️⚧️ | In reply to @jade_:matrix.org
you can use --keep-failed and it will keep the build-dir that failed and give you more meson logs, but...
this is really weird that it's consistently broken on your machine and on gha and that not really anyone else seems to be hitting it
I'm gonna try bisecting from 2.91 | 06:30:56 |
aloisw | In reply to @k900:0upti.me https://github.com/NixOS/nix/issues/11503 Why am I not surprised that the userspace chroot does not actually act like a chroot. | 08:24:48 |
raitobezarius | it's in my todolist to deal with the buildbot stuff today | 10:03:15 |
V. 🏳️⚧️ | Redacted or Malformed Event | 11:42:48 |
V. 🏳️⚧️ | Redacted or Malformed Event | 11:43:14 |
V. 🏳️⚧️ | Redacted or Malformed Event | 11:45:10 |
V. 🏳️⚧️ | Redacted or Malformed Event | 11:52:27 |
V. 🏳️⚧️ | In reply to @vigress9:matrix.org I'm gonna try bisecting from 2.91 Nvm I'm cursed, they fail on 2.91 too | 13:11:35 |
V. 🏳️⚧️ | Both on GHA and locally | 13:11:51 |
aloisw | Locally as in the devshell, or even if you build them using nix? | 13:12:15 |
V. 🏳️⚧️ | In the devshell | 13:13:12 |
V. 🏳️⚧️ | I don't think I'll get a different result if I nix build locally, since a nix build on GHA fails | 13:13:39 |
aloisw | Do you have special options in your nix.conf ? IIRC the global configuration leaks into the tests. | 13:14:01 |
aloisw | In reply to @vigress9:matrix.org I don't think I'll get a different result if I nix build locally, since a nix build on GHA fails Is the GHA repository public? | 13:14:28 |
V. 🏳️⚧️ | https://github.com/vigress8/lix-snapshot/blob/main/.github/workflows/build-lix.yml | 13:15:07 |
V. 🏳️⚧️ | In reply to @aloisw:kde.org Do you have special options in your nix.conf ? IIRC the global configuration leaks into the tests. When I do the environment variable fix to make it ignore global config, the pipe-operator test works, but the rest of the failures are unaffected | 13:17:59 |
aloisw | What operating system and filesystem is your local setup? | 13:35:09 |
V. 🏳️⚧️ | In reply to @aloisw:kde.org What operating system and filesystem is your local setup? Xubuntu 24.04, ext4 | 13:53:20 |
aloisw | I can reproduce the issue on that system, let's see if I can get somewhere. | 16:02:18 |
aloisw | Also in general it's better to not delete messages willy-nilly. | 16:03:14 |