[alsa] Bump to v1.2.11

[gnadt]

Now just bumping alsa

[gnadt]

Shoot I also just bumped [alsa_plugins] to v1.2.7, forgetting that would get added in here

[imciner2]

It's slightly awkward to hardcode them like this, but I guess there isn't really a good way around that. Have you tested this locally to make sure the paths all work out?

[giordano]

Would that work on nix?

[imciner2]

Probably not on nix, and this will also reference the user's existing Alsa stuff, not anything we provide in the JLL. There are apparently environment variables that could be used to work around this at runtime (they basically override this setting apparently).

[giordano]

I was moderately sure we were doing this already

[giordano]

#1432

[giordano]

I think it'd be better to having an init block like (it's a sketch, totally untested):

get!(ENV, "ALSA_PLUGIN_DIR", joinpath(artifact_dir, "lib", "alsa-lib"))

rather than hardcoding absolute paths that we can't know whether will exist. Other projects seem to be doing this already, e.g. JuliaMultimedia/SimpleDirectMediaLayer.jl#65

[gnadt]

Thanks for the discussion & link!

[giordano]

I think you still want to add the init block? After testing it of course 🙂 I have no idea of what to test

[gnadt]

Oh sorry I got confused, I thought the move was just to add it^ to any downstream packages using alsa_jll. I can build locally (though that's the same as the checks here), but I don't have a linux system to test locally

[giordano]

No, I was suggesting setting the keyword argument init_block to build_tarballs in this package (but pointing only to files within the tarball), so that all downstream packages automatically get the fix.

[gnadt]

I don't think I have the expertise to do that, and that's just an experimental argument, right? I'm clueless how artifact_dir would be determined during the build, since isn't that what's being built? I'm probably missing some fundamental understanding here