Haskell shell completion #106591
Haskell shell completion #106591
Conversation
| bashCompDir="$out/share/bash-completion/completions" | ||
| zshCompDir="$out/share/zsh/vendor-completions" | ||
| fishCompDir="$out/share/fish/vendor_completions.d" | ||
| bashCompDir="''${!outputBin}/share/bash-completion/completions" |
There was a problem hiding this comment.
What does ${!outputBin} do here? Is it more appropriate than ${out}?
There was a problem hiding this comment.
$outputBin is "bin" or "out".
${!x} is a double lookup of x, so ${!outputBin} evaluates to either $bin or $out.
There was a problem hiding this comment.
I am sorry, that I ask stupid questions here, but I am curious about that variable and the manual didn’t enlighten me.
I know what $out does, but how would $bin look like in which kind of derivations is ${outputBin} = $bin?
For my uninformed eyes the $bin/bin/... in this change looks like a potential bug.
There was a problem hiding this comment.
To clarify: My naive expectations would be that $bin = $out/bin
There was a problem hiding this comment.
$bin is just the bin output store path e.g /nix/store/...-niv-1.0-bin, $bin/bin is the "FHS" bin directory inside the separate bin output /nix/store/...-niv-1.0-bin/bin.
I'll also point out that the bin output isn't limited to having only a bin directory. It makes sense to have supporting files like this in the same output. A data or share output is not helpful because it's only used in conjunction with bin, negating potential closure size benefits and doing so creates a risk of creating a reference cycle in the output, which is not allowed.
There was a problem hiding this comment.
Could you rebase this on the haskell-updates branch?
Also, is this related at all to NixOS/cabal2nix#474 by @zimbatm?
roberth
force-pushed
the
haskell-shell-completion
branch
from
507c759
to
2325cd5
Compare
Done.
It has the same goal, but those packages don't use |
|
Thanks!
I'm not sure of the full backstory, but I think the idea is that if you put overrides directly into This used to be important when there were tools like |
|
@cdepillabout I had a discussion about that with peti last friday. It's exactly for the reason you say. We are brainstorming right now about redesigning the process (something which we really want to hear your opinion about, btw. we‘d like to schedule a meeting as soon as we get to it), one longterm goal might be to migrate all those overrides into nixpkgs and having tooling that can generate derivations will taking overrides defined in nixpkgs into account. |
That's probably the best option. We have been using that approached in the ruby packages and it has been working quite well. The biggest advantage is that the haskell packages can be fixed, without having to re-release cabal2nix. |
This sounds exactly like how I expected it to work. 👍 |
To be honest, I don't really have strong opinions either way. It seems like a lot of people just end up using
I agree that this does sound nice, but maybe another solution would be to add more people as maintainers for |
Motivation for this change
Make shell completion work for more haskell commands.
nivandormolurequire the "support separate bin" commit. Otherwise, it doesn't work whenniv.binis used, which is howpkgs.nivis defined (as opposed tohaskellPackages.niv).Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)nix path-info -Sbefore and after)