| 23 Nov 2024 |
matthewcroughan | Anyone got experience with fortran/medfile/ints ? | 11:47:17 |
matthewcroughan | I wanted to upstream https://github.com/MatthewCroughan/codedisaster into nixpkgs, but need some help making it happen, if anyone is interested in finite-element-analysis, but this is not related to reducing hydra failures, so only if you're interested :) | 11:48:09 |
Johannes Kirschbauer @hsjobeki | In reply to @erictapen:chat.ccc-p.org I maintain python3Packages.python-redis-test und it fails to build on darwin. I have no access to the platform. Should I just restrict the platforms then, even though upstream doesn't explicitly not support darwin? Or should I wait until end of ZHF for it? Did you look at the build log. Sometimes it is trivial to fix even if you dont have a builder | 11:56:26 |
Weijia | https://hydra.nixos.org/build/278421390/nixlog/2 | 11:57:52 |
Weijia | Probably shouldn't run the tests on darwin... | 11:57:56 |
Johannes Kirschbauer @hsjobeki | Hm it seems like they are checking the redis database connection ?
Questionable If the database tests failing if the thing itself works. | 12:00:41 |
tricktron | Kerstin (she/her): All the tests in test_redis_lock fail with AssertionError: Waited 2.00secs but ['ready to accept connections'] did not...
If you disable the test_redis_lock tests then it builds on darwin so I would just disable the tests on darwin if all tests pass on linux.
| 12:08:32 |
matthewcroughan | https://github.com/NixOS/nixpkgs/pull/296088 Koen (SynQ) de Jonge | 12:12:25 |
Lorenz | Anyone looking at Rippled? Cannot figure out why fatal: detected dubious ownership in repository at '/nix/store/hds6r54l7n6mp65gbpys50568dz52s22-NuDB/.git' | 12:24:30 |
matthewcroughan | In reply to @apeioo:matrix.org Anyone looking at Rippled? Cannot figure out why fatal: detected dubious ownership in repository at '/nix/store/hds6r54l7n6mp65gbpys50568dz52s22-NuDB/.git' Why is there a .git in the nix store? | 12:25:01 |
matthewcroughan | That is the root of the issue. Does the derivation create a .git with git init during its build? Does it have git in the nativeBuildInputs? | 12:25:21 |
Lorenz | The package specifies leaveDotGit = true; I do not know why. But the error does not change if I remove that line. | 12:26:02 |
Lorenz | * The package specifies leaveDotGit = true; in fetchgit. I do not know why. But the error does not change if I remove that line. | 12:26:30 |
Lorenz | For some reason I think the package requires a Git repository during build: git config --global url."file://${nudb}".insteadOf "${nudb.url}" | 12:29:52 |
Lorenz | * I think for some reason the package requires a Git repository during build: git config --global url."file://${nudb}".insteadOf "${nudb.url}" | 12:30:09 |
Lorenz | * It seems the build step clones some Git repositories and these are "prefetched" and configured to point to the store: git config --global url."file://${nudb}".insteadOf "${nudb.url}" | 12:33:27 |
| anothertobi joined the room. | 13:17:25 |
| anothertobi set a profile picture. | 13:21:54 |
infinisil | diff --git a/nix/store/w21dd3v32i13bg3c4nahbdd444587gdb-source/thirdparty/JUCE/modules/juce_graphics/native/juce_freetype_Fonts.cpp b/./thirdparty/JUCE/modules/juce_graphics/native/juce_freetype_Fonts.cpp
index a813ec6..7e29f3a 100644
--- a/nix/store/w21dd3v32i13bg3c4nahbdd444587gdb-source/thirdparty/JUCE/modules/juce_graphics/native/juce_freetype_Fonts.cpp
+++ b/./thirdparty/JUCE/modules/juce_graphics/native/juce_freetype_Fonts.cpp
@@ -353,8 +353,8 @@ private:
bool getGlyphShape (Path& destShape, const FT_Outline& outline, const float scaleX)
{
const float scaleY = -scaleX;
- const short* const contours = outline.contours;
- const char* const tags = outline.tags;
+ const short* const contours = (short*) outline.contours;^M
+ const char* const tags = (char*) outline.tags;^M
const FT_Vector* const points = outline.points;
for (int c = 0; c < outline.n_contours; ++c)
| 13:27:26 |
matthewcroughan | In reply to @apeioo:matrix.org The package specifies leaveDotGit = true; in fetchgit. I do not know why. But the error does not change if I remove that line. leaveDotGit is generally an issue anyway, as it can cause fetching to be non-deterministic | 13:28:34 |
matthewcroughan | if it still works without it, might be a good idea to remove it | 13:28:41 |
Lorenz | No it does not work without. It seems Git got stricter or something changed how permissions are handled in the Nix sandbox. It of course works with --add safe.directory, but I do not know why it broke. | 13:31:13 |
Corsin Pfister | Can someone add the labes to https://github.com/NixOS/nixpkgs/pull/358431 ? Thanks! | 13:38:13 |
Koen (SynQ) de Jonge | I am setting up 'test vm' there is one available at: 95.215.187.101
you can login as alice password is the same as the wifi password | 13:45:14 |
Koen (SynQ) de Jonge | * I am setting up 'test vm' there is one available at: 95.215.187.102
you can login as alice password is the same as the wifi password | 13:45:46 |
Koen (SynQ) de Jonge | * I am setting up 'test vm' there is one available at:
nix1: 95.215.187.101
nix2: 95.215.187.102
you can login as alice password is the same as the wifi password | 13:46:26 |
Koen (SynQ) de Jonge | there is tons of cpu and ram in there, and it is a 24.11 install | 13:46:50 |
Koen (SynQ) de Jonge | have a go | 13:46:53 |
Koen (SynQ) de Jonge | Oh and please claim the use by saying here 'I am using this one'. | 13:47:16 |
Marc Jakobi | another fix to my remote builders in case it was someone here was able to merge my previous one 🙏 | 13:48:13 |