Right now, the strategy the LSP server's flake uses has been copied into the overrides.nix so that the LSP server in the package set builds a functional executable. Instead, I think it would be better to treat the LSP server like Idris2 is treated: Update a flake.nix input to refer to the correct commit of the LSP repo for the given package set when running update.sh and then use that flake input by forcing it into the package set over the top of the existing idris2-lsp entry (or, for non-flake builds, pull the LSP explicitly based on the flake.lock entry).
mattpolzin
commented
2 months ago
Right now, the strategy the LSP server's flake uses has been copied into the
overrides.nixso that the LSP server in the package set builds a functional executable. Instead, I think it would be better to treat the LSP server like Idris2 is treated: Update a flake.nix input to refer to the correct commit of the LSP repo for the given package set when runningupdate.shand then use that flake input by forcing it into the package set over the top of the existingidris2-lspentry (or, for non-flake builds, pull the LSP explicitly based on the flake.lock entry).