Closed GaetanLepage closed 1 month ago
Probably also worth noting in the README how one can --override-input neovim-src, which is useful for developing against the neovim source code.
❮ nix flake check
error: builder for '/nix/store/lppg8w1ldhrsdhywg00x2v2jiin9j9pc-shlint.drv' failed with exit code 2;
last 1 log lines:
> make: *** No rule to make target 'shlint'. Stop.
For full logs, run 'nix log /nix/store/lppg8w1ldhrsdhywg00x2v2jiin9j9pc-shlint.drv'.
@willruggiano any idea ?
Are you guys fine with using alejandra
as the formatter ?
❮ nix flake check error: builder for '/nix/store/lppg8w1ldhrsdhywg00x2v2jiin9j9pc-shlint.drv' failed with exit code 2; last 1 log lines: > make: *** No rule to make target 'shlint'. Stop. For full logs, run 'nix log /nix/store/lppg8w1ldhrsdhywg00x2v2jiin9j9pc-shlint.drv'.
@willruggiano any idea ?
Looks like upstream removed it.
re shlint, looks like these targets are the ones to use now: https://github.com/neovim/neovim/blob/01b4da65c229f05ccb26c55db4e0d30ed9bac10b/Makefile#L119
Although I don't particularly see the value in having those as flake checks? It's an extra burden on the maintainers of this repo to ensure they're in sync with upstream. Also I guess it kinda depends if we want the nightlies to only include upstream nightlies that pass these checks. Could go either way honestly
re shlint, looks like these targets are the ones to use now: https://github.com/neovim/neovim/blob/01b4da65c229f05ccb26c55db4e0d30ed9bac10b/Makefile#L119
Although I don't particularly see the value in having those as flake checks? It's an extra burden on the maintainers of this repo to ensure they're in sync with upstream. Also I guess it kinda depends if we want the nightlies to only include upstream nightlies that pass these checks. Could go either way honestly
@teto , what's your opinion on this ?
Should we drop the checks
altogether ?
The references of neovim-nightly
in the README should also be changed for neovim
right ?
neovim-nightly-overlay on flake
❮ nix build --show-trace
error:
… while calling the 'derivationStrict' builtin
at /builtin/derivation.nix:9:12: (source not available)
… while evaluating derivation 'neovim-unwrapped-nightly'
whose name attribute is located at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/pkgs/stdenv/generic/make-derivation.nix:331:7
… while evaluating attribute 'preConfigure' of derivation 'neovim-unwrapped-nightly'
at /nix/store/8g5455rmasxdrkas1b8f8ycscjavvcp1-source/flake/packages/neovim.nix:86:5:
85| inherit src;
86| preConfigure = ''
| ^
87| ${oa.preConfigure}
… while calling anonymous lambda
at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/lib/attrsets.nix:1096:10:
1095| attrs:
1096| map (name: f name attrs.${name}) (attrNames attrs);
| ^
1097|
… from call site
at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/lib/attrsets.nix:1096:16:
1095| attrs:
1096| map (name: f name attrs.${name}) (attrNames attrs);
| ^
1097|
… while calling anonymous lambda
at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/pkgs/by-name/ne/neovim-unwrapped/package.nix:172:18:
171| '' + lib.concatStrings (lib.mapAttrsToList
172| (language: src: ''
| ^
173| ln -s \
… while evaluating derivation 'bash-grammar-neovim-nightly'
whose name attribute is located at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/pkgs/stdenv/generic/make-derivation.nix:331:7
… while evaluating attribute 'src' of derivation 'bash-grammar-neovim-nightly'
at /nix/store/0kl2p60mvnpscnza21p8583sz3f60jxf-source/pkgs/by-name/ne/neovim-unwrapped/package.nix:175:20:
174| ${tree-sitter.buildGrammar {
175| inherit language src;
| ^
176| version = "neovim-${finalAttrs.version}";
error: cannot coerce a set to a string
Hmmmm
look at my PR, that's because neovim expects src =
so you need to convert the generated grammar sets into a src= fetchurl
attribute entry
look at my PR, that's because neovim expects
src =
so you need to convert the generated grammar sets into asrc= fetchurl
attribute entry
Sort of. That is how it worked a few days ago, but that commit has since been reverted; https://github.com/NixOS/nixpkgs/commit/76ef4c7888c52bd4eed566011c24da9eb437a3c8
damn. We could pin nixpkgs to an earlier version then, the code is going to be re-reverted again after branch-off . We are more interested in neovim bump than nixpkgs-bumps anyway
damn. We could pin nixpkgs to an earlier version then, the code is going to be re-reverted again after branch-off . We are more interested in neovim bump than nixpkgs-bumps anyway
Yeah, we should also (temporarily) prevent hercules-ci from auto updating that input. No idea how to do that though lol.
EDIT: looks like hercules-ci.flake-update.flakes.".".inputs = [ ...everything but nixpkgs for now... ];
should do the trick
The neovim update PR has been re-reverted, so the current neovim version in nixpkgs
is now 0.10.0 again.
Should I still pin nixpkgs ?
it's a bit peculiar to have everything scattered across multiple files but that's an implementation detail. It seems to work. I wonder why the CI check is waiting ? @kranzes is a server unreachable ? I will merge anyway, the neovim core team is impatient to get rid of the flake and the basics seems to work, we can refine the details later. Thanks to everyone involved
@Kranzes seems like I can't force a merge without CI. I have limited access to the repository page. Can you merge this/fix CI ? also maybe invite @willruggiano and @GaetanLepage if you agree ?
Adding y'all now...
@Kranzes is the CI doing nix flake check
? or more ? or less ?
I added @GaetanLepage but wasn't able to add @willruggiano as they are not part of the nix-community org, can you ask for an invite on the nix-community matrix room? cheers.
@Kranzes seems like I can't force a merge without CI. I have limited access to the repository page. Can you merge this/fix CI ? also maybe invite @willruggiano and @GaetanLepage if you agree ?
Hercules-CI currently doesn't support running on forks, you could use some GitHub feature/bot to pull it into a different branch on this repo so it runs, but I haven't needed to figure that out yet. I suggest for further development here you should use your own branch here rather than a fork.
@Kranzes is the CI doing
nix flake check
? or more ? or less ?
The CI builds all of the outputs and pushes them to the binary cache, it also runs nix flake update every midnight and makes a PR for it and auto-merges it if it builds for Linux.
This pull request has been mentioned on NixOS Discourse. There might be relevant details there:
https://discourse.nixos.org/t/neovims-flake-nix-moved-to-nix-community/45887/1
I added @GaetanLepage but wasn't able to add @willruggiano as they are not part of the nix-community org, can you ask for an invite on the nix-community matrix room? cheers.
@Kranzes this one? #nix-community:nixos.org
had this in my home.nix
which doesnt work now:
nixpkgs.overlays = [
(import (builtins.fetchTarball {
url = https://github.com/nix-community/neovim-nightly-overlay/archive/master.tar.gz;
}))
];
is there anything to change here? an update to readme would be nice
@Kranzes this one? #nix-community:nixos.org
This one: https://matrix.to/#/#neovim-nightly-overlay:nixos.org
This one: https://matrix.to/#/#neovim-nightly-overlay:nixos.org
Were you able to join @willruggiano ?
Maintaing the flake in the official neovim repository became too much of a burden so we are moving it out to a nix-community repository.
This is the opportunity to do more and add automation so that the neovim nix overlay doesnt ever break !
Follow up of #515