nix-output-monitor: Add nom-build script #114903
Conversation
|
I am torn on the question if this is something I should put in the upstream repository. But patching the paths will be at least the same number of lines here in post-install, so I don‘t see a big benefit. |
| @@ -25,6 +26,10 @@ mkDerivation { | |||
| ansi-terminal async attoparsec base containers directory HUnit mtl | |||
| nix-derivation process relude stm text time unix | |||
| ]; | |||
| postInstall = '' | |||
| echo -e "#!/bin/sh\n${expect}/bin/unbuffer nix-build \"\$@\" 2>&1 | exec $out/bin/nom" > $out/bin/nom-build | |||
There was a problem hiding this comment.
| echo -e "#!/bin/sh\n${expect}/bin/unbuffer nix-build \"\$@\" 2>&1 | exec $out/bin/nom" > $out/bin/nom-build | |
| echo -e "#!/bin/sh | |
| ${expect}/bin/unbuffer nix-build \"\$@\" |& $out/bin/nom | |
| " > $out/bin/nom-build |
Why do we want exec here?
There was a problem hiding this comment.
idk, I don‘t know of a disadvantage and I don‘t see that we need the running bash process after nom has been started. Also maybe exit code propagation?
There was a problem hiding this comment.
exec is really just to get rid of the bash process, there’s no reason to keep it around. It also will keep the ps output cleaner.
There was a problem hiding this comment.
This would probably also be a bit cleaner if written using a here-doc. At a bare minimum, I don’t see why this is using the -e flag to echo. And the shebang should probably be #!${runtimeShell}.
There was a problem hiding this comment.
Heck, you could probably make this a writeShellScript instead, using the outPath attribute of the resulting derivation, like
let
nom-build = writeShellScript "nom-build" ''
${expect}/bin/unbuffer nix-build "$@" |& ${self}/bin/nom
'';
self = mkDerivation rec {
# …
postInstall = ''
install -Dm755 ${nom-build} $out/bin/nom-build
'';
};
in selfI haven’t tried this directly myself, but I think the outPath is available without forcing anything beyond name and fixed-hash attrs.
There was a problem hiding this comment.
If thought about that, but was un-optimistic. I’ll try.
There was a problem hiding this comment.
No, that doesn‘t make sense. If the hash of nom changed the hash of nom-build needs to change and vice versa. I‘ll try anyways just to make sure, but I would be thoroughly surprised.
There was a problem hiding this comment.
Thanks for the feedback, I have rewritten the function with your hints. Looks much better now.
|
Now we need bash-completion for nom-build. |
Fair enough. But I have no clue how to do that … |
|
|
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/announcing-the-nix-output-monitor/11672/7 |
Motivation for this change
By popular demand we present
nom-buildit's exactly likenix-buildbut with morenom.Credits to @lilyball for the idea:
https://discourse.nixos.org/t/announcing-the-nix-output-monitor/11672/6
My bash in nixpkgs foo is not very strong, so I appreciate any style critique.
\cc @mweinelt
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)nix path-info -Sbefore and after)