pkgsMusl.jdk8: mark broken for musl

[superherointj]

pkgsMusl.jdk8: mark broken for musl

Tracking issue: #281618

error: ‘isnanf’ was not declared in this scope

CC @edwtjo

[superherointj]

I'm not sure what is going on with CI:

[wegank]

@infinisil

[infinisil]

Sorry for that! Was broken with #281407, but now already fixed with #281412, it's really easy to mess up the GitHub workflows since they're not tested by default 😅. You'll have to rebase unfortunately.

[yu-re-ka]

pkgsMusl.jdk8: mark broken for musl

I'm not sure if it makes sense to start marking everything as broken on musl that doesn't build. You are going to add a lot of broken = stdenv.isMusl in that case...

[superherointj]

I'm not sure if it makes sense to start marking everything as broken on musl that doesn't build. You are going to add a lot of broken = stdenv.isMusl in that case...

Better than building it every time and finding out some inner dependency is broken. As Musl doesn't have cache, builds are longer.

Every PR I have opened to report breakage is specific to the package and not some downstream breakage. So, it is not all packages that are being marked broken but just the packages that need to be marked broken to avoid long builds.

Also, I find a chore and tiring having to do this. But someone has to do the dirty work.

Some problems have a known solution from others, so it is important to make it known what is broken, so others can chime in. For example, the php package, I got a helpful hint. And maintainers might have an interest in fixing their packages, which just happened to pony.

Once the specific broken packages are marked broken, the set of packages that are needed to be built is smaller, making build resource usage lower, hopefully enabling to cache musl. Cache is important for CI builds and to avoid regressions, such as this.

So it is a step in the right direction for the musl ecosystem to flourish.