| 17 Dec 2024 |
GGG | Would have to evaluate the impact of doing that widespread though | 22:00:02 |
6pak | this is where the fact I had the convo with Corngood in dms backfires... | 22:00:14 |
GGG | In reply to @6pak:matrix.org you should fill it out based on hostSystem No, the RuntimeIdentifiers should have all supported platforms, no? | 22:00:45 |
6pak | it should have everything you currently want to build | 22:01:06 |
6pak | and in the nixpkgs model, you can only build one at a time | 22:01:11 |
GGG | Then the `--runtime` flag tells which one were building for | 22:01:12 |
GGG | In reply to @6pak:matrix.org it should have everything you currently want to build But if we only set it to one, then it'll screw over the restore process and fetch-deps | 22:01:47 |
6pak | afaik RuntimeIdentifiers only affects nuget restore | 22:01:48 |
GGG | Yeah | 22:01:55 |
6pak | and it doesn't make sense to download stuff you don't need to build what you want | 22:01:56 |
6pak | because it's wrong imo | 22:02:18 |
GGG | Which is why I was thinking of setting it based on the `meta.supportedPlatforms` | 22:02:24 |
GGG | That way it'd only restore for the platforms it supports | 22:02:37 |
6pak | fetch-deps output should be specific for the platform | 22:02:39 |
GGG | No, I disagree with that | 22:02:51 |
6pak | because you shouldn't need to download unfree windows deps if you build a linux app | 22:02:57 |
6pak | just because it supports windows target | 22:03:01 |
GGG | Sure, we can have per platform dependencies in the file | 22:03:05 |
GGG | But the file should have dependencies for all platforms | 22:03:15 |
6pak | sure it can be in one file whatever | 22:03:20 |
GGG | Otherwise it'd be hell generating the `deps.json` file | 22:03:28 |
6pak | it shouldn't pull in dependencies for other platforms | 22:03:30 |
GGG | Sounds doable | 22:04:03 |
6pak | my issue with buildDotnetModule is that it doesn't really work well with nixpkgs model | 22:04:52 |
GGG | How so? | 22:05:00 |
6pak | where every derivation has one hostSystem | 22:05:02 |
6pak | where in dotnet you tend to have multiple projects in solutions that have different kind of outputs | 22:05:24 |
6pak | and you can't really break up projectreferences into derivations easily | 22:05:32 |
GGG | It'd also make it quite a pain to package anything | 22:05:56 |
GGG | If you had to make a derivation for every single csproj | 22:06:14 |