-
Notifications
You must be signed in to change notification settings - Fork 336
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat(shell): revert from naked shell to pkgs.mkShell + unsetting irrelevant env vars #507
Conversation
Also fixes #486. |
Ok I tested this against the flake in rules_ll and things seem to work fine. This change of course leads to more environment variables being set than before, but at least we know that it won't break existing more "exotic" C++ toolchains 😊 |
It seems that Nim is broken in nixpkgs, not here: https://hydra.nixos.org/build/213226629 Possible fix NixOS/nixpkgs#222285 |
0e7c7ed
to
7bdefce
Compare
Thanks, I removed the irrelevant change. |
Also my neovim could not find |
Although this patch addresses a lot of issues I am still seeing some issues on nix-darwin, specifically with Foundation, maybe an issue with nativebuildinputs not propagating. These vars seem to fix build, most of them are available though through this patch. So not sure:
|
Hi, sorry may I ask is there any update on this PR? |
I mainly would like people to test it thoroughly before it's merged. |
I think it will be easier for people to test if we merge after all the CI check passed. If there is any issue, we can create a PR to fix. WDYT? |
I was actually wondering: What are the benefits of unsetting all those environment variables? I didn't really get the "keep it lean" comment. |
Hi, how long do we expect people to test this out before merging? |
I've updated this branch in my fork: https://github.com/bobvanderlinden/devenv/tree/feat/stripped-shell There will be some regressions when merging this. Question is whether that is a good thing (honestly do not know). With This is why the bobvanderlinden@b669f22#diff-3ea46e5d7bf8933097511b155be31bf25b2c6e01081c6d9a6dfe61624717082aR4 Question is whether setting Maybe as an alternative we should introduce a different option. Something like The real solution is for people to not rely on pre-built binaries, but I think that's the world we live in right now. |
IMO it's risky to let users unknowningly manipulate these variables, but if it's explicit it could be a viable option. Though I wonder whether such an option is actually necessary. Personally, I think I'd prefer explicitly setting |
is it possible now to add openssl dev deps? |
Yes, the packages are added as
There are people already relying on just adding runtime libraries as packages. Having a separate option for runtimeLibraries will make it a bit clearer how to migrate. It also allows to document what it does and why it is a bad idea to use it. The specific use-cases where it's the only option is also good to mention in the docs. I agree it's a bad idea in general, but shipping binaries is the underlying problem. That's sometimes quite hard to get around when using ruby/python/node package managers. |
This seems like the right thing to do. Setting |
I've done testing today on examples on top of #745 and python-poetry example fails with
We'll need to fix this one, any ideas? I tried adding |
I've merged this into #745, but we'll need to fix the errors before it can be merged. |
Great, thanks for bringing this forward. Closing this PR. |
Thank you for making this happen ❤️ |
closes #498
fixes #497