| 7 Mar 2025 |
 lzcunt | In reply to @6pak:matrix.org
just make a regular stdenvNoCC.mkDerivation which uses mono's xbuild in buildPhase and nuget restore in configurePhase and wrap that in addNuGetDeps yep that's the plan | 15:02:03 |
| lzcunt | In reply to @sananatheskenana:matrix.org
I think the issue might be that nuget restore fetches the nupkg instead of just the metadata, and in fact the metadata is not in the output only the package is maybe we could add nuget restore support to nuget-to-json? It'd essentially be using the already fetched nupkg instead of prefetching the url from the metadata | 15:03:15 |
| lzcunt | so basically since there's no metadata in the output directory, nuget-to-json is in an infinite loop i think | 15:05:06 |
 GGG | In reply to @sananatheskenana:matrix.org
I think the issue might be that nuget restore fetches the nupkg instead of just the metadata, and in fact the metadata is not in the output only the package is No, as long as it unpacks the packages into the directory it should be fine | 15:22:31 |
| GGG | Lemme review the source, but I don't remember it depending on any specific metadata on the output directory | 15:23:28 |
| GGG | I think it might be stuck here instead: https://github.com/NixOS/nixpkgs/blob/master/pkgs%2Fby-name%2Fnu%2Fnuget-to-json%2Fnuget-to-json.sh#L26-L28 | 15:24:44 |
| lzcunt | I tested that command manually and it worked so I assumed it wasn't | 15:25:31 |
| lzcunt | IIRC the output was something like `E https://whatever` | 15:25:49 |
| lzcunt | so the second row has URLs | 15:26:06 |
| lzcunt | What does a proper packages directory look like when generated by `dotnet restore` (also how do inline codeblocks work in matrix again? | 15:27:42 |
| GGG | It should have one directory per package and then a subdirectory with the package version and then inside have the package files | 16:37:05 |
| GGG | * It should have one directory per package and then a subdirectory named with the package version and then inside have the package files | 16:37:22 |
|  Qyriad changed their display name from Qyriad to qyriad. | 16:43:54 |
| lzcunt | In reply to @gggkiller:matrix.org
It should have one directory per package and then a subdirectory named with the package version and then inside have the package files IIRC that wasn't how nuget restore generated them, it had one directory per package with a .nupkg file and sometimes other files too | 17:44:33 |
| lzcunt | But I'll test it again | 17:44:43 |
| lzcunt | * IIRC that wasn't how nuget restore generated them, it had one directory per package with a .nupkg file and sometimes other files too, no version directory | 17:45:18 |
| 8 Mar 2025 |
| Qyriad changed their display name from qyriad to Qyriad. | 21:41:03 |
| 9 Mar 2025 |
|  Gaël joined the room. | 23:05:31 |
| 12 Mar 2025 |
|  @rosssmyth:matrix.org joined the room. | 00:41:13 |
| 15 Mar 2025 |
|  galactic_starfish joined the room. | 03:01:22 |
| 17 Mar 2025 |
| GGG | well, apparently Jetbrains Rider breaks with scudo allocator | 06:01:08 |
| GGG | * well, apparently Jetbrains Rider breaks with scudo, graphene-hardened and graphene-hardened-light allocators | 06:08:47 |
| GGG | fortunately seems like it work with mimalloc with the security features enabled, at least I can have some extent of security in the allocator | 06:26:05 |
| GGG | * ~~fortunately seems like it work with mimalloc with the security features enabled, at least I can have some extent of security in the allocator~~
nevermind, chromium breaks with mimalloc | 07:22:30 |
| GGG | * ~fortunately seems like it work with mimalloc with the security features enabled, at least I can have some extent of security in the allocator~
nevermind, chromium breaks with mimalloc | 07:22:34 |
| GGG | * fortunately seems like it work with mimalloc with the security features enabled, at least I can have some extent of security in the allocator
nevermind, chromium breaks with mimalloc | 07:22:41 |
| GGG | * fortunately seems like it work with mimalloc with the security features enabled, at least I can have some extent of security in the allocator
nevermind, chromium breaks with mimalloc | 07:23:59 |
 Matt Sturgeon | Am I missing something, or is fetch-deps not working? Seems like it builds the nix shell but doesn't actually run the fetch-deps.sh script, so I don't see any changes in deps.json.
To reproduce on latest master, wipe a deps.json file and attempt to re-generate it:
$ echo "" > pkgs/by-name/ne/nexusmods-app/deps.json
$ nix-build -A nexusmods-app.fetch-deps
$ ./result
No changes to add-nuget-deps in 3months, so I'm confused. Nothing obviously wrong in the built scripts either.
| 22:46:46 |
| Matt Sturgeon | * Am I missing something, or is fetch-deps not working? Seems like it builds the nix shell but doesn't actually run the inner fetch-deps.sh script, so I don't see any changes in deps.json.
To reproduce on latest master, wipe a deps.json file and attempt to re-generate it:
$ echo "" > pkgs/by-name/ne/nexusmods-app/deps.json
$ nix-build -A nexusmods-app.fetch-deps
$ ./result
No changes to add-nuget-deps in 3months, so I'm confused. Nothing obviously wrong in the built scripts either.
| 22:47:16 |
| GGG | yeah, there's an open issue about it, curl now fails with --netrc flag and we add it unconditionally in nuget-to-json | 22:47:39 |
