nixUnstable is a deprecated alias, so the script didn't work anymore.
Also, I see no need to avoid the default nix here (maybe years ago).
Tested on mirror-tarballs.service at pluto.nixos.org (atop 24.05).
When installing a lua package that depends on luarocks, luarocks was not
capable of recognizing the luarocks module: https://github.com/NixOS/nixpkgs/issues/316009 .
As explained here https://github.com/luarocks/luarocks/issues/1659,
luarocks needs a manifest to recognize the package.
I first attempted to generate a manifest with `luarocks-admin
make-manifest` on our current manual luarocks derivation but this ended
up being too hackish as the `make-manifest` requires a .rockspec or
.rock to work, which we had not.
It seemed simpled to just rename the current manual luarocks derivation to luarocks_bootstrap
in order to break cyclic dependencies in buildLuarocksPackage and add a
generated luarocks derivation. Previously it was impossible to test
luarocks because of the cyclic dependency but we will now be able to
test the generated derivation.
- kio-extras-kf5 moved to misc because of release schedule weirdness
- calindori ported to KF6, reenabled
- some build fixes all over
- add 7zip to a bunch of games for SVG compression
This is a breaking change that converts the 'src' csv column (a git URI) to a 'rockspec'
fullpath uri.
before this, luarocks-nix would clone the repo and look for rockspecs
with a brittle algorithm. This removes the ambiguity on which rockspec
to package while allowing to remove a bunch of brittle code from
luarocks-nix.
the intention being to catch commits which declare themselves as
cherry-picks, but either:
- don't refer to a commit in the master or staging branches
- are significantly altered from their original commit
determining the latter is not an exact science, but the heuristic of
looking for differences in only the added or removed lines seems to
work quite well. still, this should be considered an assistant
for reviewers rather than a hard failure. unfortunately github
workflows don't have a way of raising a gentle warning instead of a
failure.
the formatting of the output also leaves something to be desired due
to the limitations of github actions' "group" commands.
The nixpkgs-unstable channel's programs.sqlite was used to identify
packages producing exactly one binary, and these automatically added
to their package definitions wherever possible.
