| 8 Jun 2025 |
 pveierland | I agree, it would be way easier to just do a possibly lossy conversion of all attrpaths and paths to UTF8, and it would probably result in fewer bugs overall + less code and specialized parsers etc | 13:50:15 |
| pveierland | Just trying to avoid losing any necessary representational ability | 13:51:09 |
| pveierland | Then again, if a tool like git was rebuilt today maybe it would just insist UTF8 everywhere | 13:51:36 |
| pveierland | I guess the issue for tools such as nix and git is that they need to work with such a broad set of environments so it might not be possible to make that assumption yet | 13:53:54 |
| pveierland | Redacted or Malformed Event | 13:54:30 |
 emily | Nix not doing Unicode properly is just historical reasons I think. | 15:04:07 |
| emily | it's not like it makes it handle e.g. UTF-16 better, because it can't do NUL bytes either. | 15:04:20 |
 dramforever | well, those should be url encoded, right? | 16:32:45 |
| dramforever | i guess it also breaks with non-ascii separately some other way | 16:33:47 |
 Las | John Ericsonwhat should I do to get 12668 merged, or has the equivalent work already been done? Haven’t been following development for a bit | 17:02:14 |
 John Ericson | equivalent work has not been merged | 17:06:38 |
| John Ericson | you can try doing a rebase | 17:06:46 |
| John Ericson | and also let me try to find again the rebase I had done before | 17:06:59 |
| John Ericson | Las: yeah l-as-goal-direct-io in my repo is a rebase that doesn't quite work of the first commit | 17:11:43 |
| 9 Jun 2025 |
|  @trofi:matrix.org left the room. | 05:27:17 |
| dramforever | i can't reproduce this on 2.28.3 | 05:38:22 |
| dramforever | $ nix eval '.#è'
4
$ nix eval '.#测试'
3
$ cat flake.nix
{
outputs = { self }: {
"测试" = 3;
"è" = 4;
};
}
| 05:39:26 |
| dramforever | oh wait does it have to be registry? | 05:39:30 |
| dramforever | oh yeah that's odd. | 05:40:03 |
| dramforever | * oh yeah that's odd... | 05:40:06 |
| dramforever | so the problem seems to be confined to parsing? i suspect the fix will be easy | 05:41:25 |
| dramforever | * so the problem seems to be confined to parsing flake urls? i suspect the fix will be easy | 05:41:33 |
| dramforever | oh yeah this is so funny
$ nix registry add test path:$(pwd)
$ nix eval 'test#%C3%A8'
4
$ nix eval 'test#%E6%B5%8B%E8%AF%95'
3
| 05:48:51 |
| dramforever | yeah it's broken | 05:51:44 |
| dramforever | you can't nix profile install it | 05:51:54 |
| dramforever | * you can't nix profile install it, for example | 05:52:06 |
| dramforever | but i still think it's not fundamentally unfixable | 05:52:15 |
| dramforever | unrelatedly
John Ericson i just noticed this meeting minutes https://discourse.nixos.org/t/2025-05-28-nix-team-meeting-minutes-229/65205 especially about #13042, and ... it doesn't really make sense to me.
Appears to be a problem with running 32-bit x86 on x86_64-linux. This should work, but is possibly host-dependent and seccomp may interfere.
No, it's trying to run 64-bit programs in a 32-bit derivation, which fails as expected. It should not work and the build is wrong.
The basic issue is that the code was written assuming cross
Incomplete sentence? If "assuming cross builds don't run tests" was meant then yes that's correct
(Temporarily dropping this here because I'm not sure where I should be putting it.)
| 06:46:19 |
| 10 Jun 2025 |
|  Danny joined the room. | 18:10:07 |
| 11 Jun 2025 |
| Las | Maybe I'm dumb, but I did git clone ... ; nix develop .#native-clangStdenv ; meson build and it fails now? | 13:54:32 |
