Conversation
710fd9d to
6b70b79
Compare
|
Perhaps needs more checking |
|
Potentially controversial PR #130608 |
|
I think that sums up pretty much the last round of PRs. There are some more changes coming, but there shouldn't be that many more PRs coming. Last but not least, I'm happy to revert any of those PR if you have strong feelings against any of them. |
|
Controversial PR #130621 |
|
These should target staging right? |
|
@Artturin Thank you for your comment. I tried to explain in the PR. If I target staging, I'll probably never get ofborg to evaluate. As a hack I'm targeting master to get faster evaluation. Once people are happy, I will contact someone from the hydra team and will target staging. I'm still not sure this is the best way to proceed, I've seen other people doing it and it seems to work well. |
okay 👍 i didn't see that comment |
pkgs/stdenv/generic/setup.sh
Outdated
There was a problem hiding this comment.
Should that be an and not or?
There was a problem hiding this comment.
I did mess it up! :-)
There was a problem hiding this comment.
I was really split between making it >1 || <-1 or keeping it like the original <= 1 && >= -1
(in the end I just mixed them up)
6b70b79 to
34bef33
Compare
|
This PR has to be tested thoroughly as I'm not sure what is happening with strings containing double @ signs in them. I'm still "practicing" here and hope to get in the deep later. |
|
semi-controversial ? |
|
makes something a little more verbose |
|
This is the last PR in this series for now Hopefully I can merge some of those. |
I'm not aware. It seems quite probable that there's no reason against and noone's dared/bothered to look into switching the default so far. (introduced in PR #53628) |
|
I'm trying an upgrade to bash 5 by default, let's see what the CI says |
|
@vcunat Just a friendly ping about testing a group of commit. |
34bef33 to
f225100
Compare
f225100 to
3c3c123
Compare
3c3c123 to
a3d67cb
Compare
|
hydra has passed, this PR is going to staging. |
Motivation for this change
Instead of using an intermediary variable, use a named reference.
the named reference feature was introduced in bash 4.3 and is exactly the use case.
IMO it is an improvement, however I want to check if others feel that way toward using a named reference (there are other places where this can be used).
I had a PR previously to enable some stdenv "improvements" that actually broke the build (#127736)
The idea is to try to break that PR logically into commits so things can be easily accepted or rejected.
Some of the changes will be related to enabling another shell, some of the changes will be things that I think should be improvements.
I will try to mark each PR as "Can be controversial" vs "I think this is OK". The current PR IMO is "I think this is OK"
I will cc people that I think would be interested in this PR and just add in the comments the other PRs that are related.
@aszlig @matthewbauer
If you know of somebody else that could be interested, I would be happy to know.
Please note I'm only using master as a faster way to get some evaluations. My plan is to first break down the previous big PR into separate smaller ones. Then separately build on my machine the ones that don't fail the CI. Then ask someone from the hydra team to test them. (as the previous one broke quite some stuff and caused pain).
I also have some questions if anybody knows the answer.
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)