| 7 Nov 2025 |
Corngood | If it's working, fetch-deps shouldn't add it at all. Have you tried setting the package and reference version both to just 1.0.0? | 02:55:35 |
Emma [it/its] | well the issue is that i want to publish the resulting files to nuget | 02:56:26 |
Emma [it/its] | i'd assume that if i made it 1.0.0, it wouldnt link up correctly in nuget, no? | 02:56:42 |
Emma [it/its] | why is dotnet-vmr so large 😔 | 03:10:19 |
Emma [it/its] |  Download clipboard.png | 03:10:34 |
Emma [it/its] | i made the mistake of purging my nix store | 03:10:46 |
Emma [it/its] | well there goes 6min45s of my life just waiting for dotnet-vmr to download | 03:12:07 |
Corngood | In reply to @emma:rory.gay i'd assume that if i made it 1.0.0, it wouldnt link up correctly in nuget, no? Yeah just trying to understand what's going on a bit better | 03:12:39 |
Emma [it/its] | fwiw my current (seemingly working) workaround is uh... | 03:13:04 |
Emma [it/its] | ArcaneLibs = arcanelibs.packages."${pkgs.stdenv.hostPlatform.system}".ArcaneLibs.overrideAttrs (old: {
__intentionallyOverridingVersion = true;
version = builtins.substring 0 29 old.version; # "1.0.0-preview-20251106-123456";
});
| 03:13:20 |
Corngood | And with that fetch-deps doesn't add anything? | 03:15:18 |
Corngood | Is it just because it happens to be newer than anything on nuget.org? | 03:15:50 |
Emma [it/its] | oh that, no that's unrelated | 03:16:06 |
Emma [it/its] | dotnet thinks the unlisted package is "newer" due to semver (old format being 1.0.0-preview987654321.abcdefg, new being 1.0.0-preview.20251106-123456+abcdefg, so a lexical string comparison would prefer the former) | 03:18:41 |
Emma [it/its] | and well, its nuget.org, so theres nothing i can do about that, no way to delete old versions or whatever | 03:20:20 |
Corngood | I still think the PackageReference version should just be pinned to the one from nix, and if that's not working because of something peculiar with the version number, trying 1.0.0 would prove it. | 03:24:22 |
Corngood | You want to stop it from looking in nuget.org at all, so either an exact version match or package source mapping. | 03:25:21 |
Emma [it/its] | that sounds like pain maintenance wise ^^'
i think ill stick to manually removing it from deps.json for the time being? | 03:26:43 |
Emma [it/its] | in other news, since you maintain dotnet-vmr in nixpkgs: any reason why the build output contains tarballs of the {dotnet,aspnetcore}-runtime packages? i'm not too familiar with building dotnet from source | 03:27:50 |
Emma [it/its] | (particularly in the lib/Private.SourceBuilt.Artifacts.10.0.100-rc.2.25502.107.linux-x64/ directory) | 03:28:44 |
Corngood | In reply to @emma:rory.gay that sounds like pain maintenance wise ^^'
i think ill stick to manually removing it from deps.json for the time being? It shouldn't be that big of a deal. You could for example set a package source mapping in preConfigure | 04:19:56 |
Emma [it/its] | oh? how would i do that? | 04:20:40 |
Corngood | In reply to @emma:rory.gay in other news, since you maintain dotnet-vmr in nixpkgs: any reason why the build output contains tarballs of the {dotnet,aspnetcore}-runtime packages? i'm not too familiar with building dotnet from source The vmr package shouldn't really be used directly if that's what you mean. Generally you want the runtime/sdk packages, which should have to references to it | 04:21:05 |
Emma [it/its] | well yeah, im using the sdk package | 04:21:18 |
Emma [it/its] | i was just surprised by the 2.4GB download of dotnet-vmr ^^' | 04:21:29 |
Corngood | How is it being referenced? You shouldn't need it for the ask | 04:21:59 |
Corngood | * How is it being referenced? You shouldn't need it for the sdk | 04:22:07 |
Emma [it/its] | if nix-output-monitor isnt lying to me, it was for dotnetHook | 04:22:28 |
Corngood | In reply to @emma:rory.gay oh? how would i do that? Add it to nuget.config for the _nix source (I'm pretty sure that's what it's called. Look at the source mapping docs | 04:23:00 |
Corngood | If anything has runtime deps on vmr that's bad. We should add some asserts for that | 04:23:46 |