| 2 Nov 2025 |
 szlend | I think it also broke tests for this thing: https://github.com/NixOS/nixpkgs/pull/359214 | 12:58:11 |
 emily | maybe we can upstream a fix like "if env did not have that thing set in its exec environment, did not have it explicitly set on the CLI, and has it now, unset it before calling exec" | 13:21:03 |
| emily | unless this is just SBCL linking against CoreFoundation and not related to env | 13:22:03 |
| emily | I'm wondering about the ~/.CFUserTextEncoding thing though. build users shouldn't have that file, surely? | 13:22:27 |
| emily | I wonder if applying https://github.com/NixOS/nixpkgs/pull/457703 fixes it | 13:24:19 |
| emily | if it does we should probably consider ^ | 13:24:29 |
 Sarah Clark | I need a bit of help. I'm working on python3Packages.bloodyad; it builds fine on linux but builds fail in the installPhase on Darwin with FileExistsError: File already exists: /nix/store/k8nji11liriv7ipwnz8wif5ajnlw1fxp-python3.13-bloodyad-2.5.0/bin/blood… | 22:14:39 |
| Sarah Clark | Happens even after a garbage collect, where I confirm the file doesn't exist before the build | 22:15:12 |
| Sarah Clark | The exported module is bloodyAD; I suspect we're hitting an issue with the case-insensitive filesystem | 22:15:42 |
| Sarah Clark | Redacted or Malformed Event | 22:15:43 |
| Sarah Clark | https://github.com/NixOS/nixpkgs/pull/457902 | 22:16:06 |
| Sarah Clark | Or do I perhaps just have to mark it bad on Darwin? | 22:16:28 |
 Randy Eckenrode | Wish we had a case sensitive store by default. Unfortunately, there’s no way to convert existing stores. | 22:33:25 |
 samasaur | maybe i'm misunderstanding but isn't this the opposite of the problem? like (assuming genv is coreutils env) genv -i env properly prints nothing, which implies to me that genv is clearing the environment before calling exec, but env -i genv prints __CF_USER_TEXT_ENCODING | 22:39:17 |
| samasaur | like doesn't that imply that the fix would be in cases where env doesn't already clear the environment? | 22:40:51 |
| samasaur | * like doesn't that imply that the fix would have to be in cases where env doesn't already clear the environment? | 22:40:55 |
| Sarah Clark | I found it after a bisect --they added the second line here:
[project.scripts]
bloodyAD = "bloodyAD.main:main"
bloodyad = "bloodyAD.main:main"
| 22:43:05 |
| Randy Eckenrode | Are they trying to transition the script name to a different casing? Why would they do that? | 22:49:04 |
| Sarah Clark | Yes, trying to make it case-insensitive. old: bloodyAD, new: bloodyad | 22:50:12 |
| Sarah Clark | Anyway, dropping line 2 above using substituteInPlace worked | 22:50:56 |
| Sarah Clark | No scripts in nixpkgs use the mixed-case form | 22:51:15 |
| 3 Nov 2025 |
 ivy | i got a pr current in progress to allow linking paths with spaces in them | 08:49:48 |
| ivy | which would finally allow linking Application Support nicely | 08:49:57 |
| ivy | as currently
buildEnv {
pathsToLink = ["/Library/Application Support"];
}
| 08:50:36 |
| ivy | does not work | 08:50:39 |
| ivy | its in staging-next now | 08:50:43 |
| ivy | and should after that be in master | 08:50:50 |
| 4 Nov 2025 |
|  yamabukiiro joined the room. | 06:18:42 |
| 5 Nov 2025 |
|  Isolaryn joined the room. | 02:23:42 |
 alexfmpe | $ echo "stdenv.buildPlatform.config" | nix repl --file . --argstr system "x86_64-darwin"
"x86_64-apple-darwin"
I don't have remote builders so I guess the fact I can build stuff this way means I'm using rosetta? I do see test suites running
| 20:54:12 |
