Skip to content
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

Staging next #97146

Merged
merged 205 commits into from Sep 7, 2020
Merged

Staging next #97146

merged 205 commits into from Sep 7, 2020

Conversation

@FRidh
Copy link
Member

@FRidh FRidh commented Sep 4, 2020

Motivation for this change
Things done
  • Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
  • Built on platform(s)
    • NixOS
    • macOS
    • other Linux distributions
  • Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
  • Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
  • Tested execution of all binary files (usually in ./result/bin/)
  • Determined the impact on package closure size (by running nix path-info -S before and after)
  • Ensured that relevant documentation is up to date
  • Fits CONTRIBUTING.md.
B4dM4n and others added 30 commits May 7, 2020
This came up in #96008.
Without this patch modinfo always prints the module name for builtin
modules, even if an explicit --field is passed.
The default `strip` invocation tries to strip global symbols from the
library, and refuses because those are indirect symbol table
references.
The latter doesn't exist on Darwin (and, presumably, other targets as
well). This change allows emacsGcc from emacs-overlay to attempt to
build where previously, the derivation would error out.
@FRidh
Copy link
Member Author

@FRidh FRidh commented Sep 5, 2020

Darwin-stdenv does not build #95910 (comment) because of texinfo change. Proposed revert #97218.

We apparently didn't fit anymore.  I don't think this test is meant
to (also) check closure size.

Note: as of this commit, the test is blocked by a fontconfig problem,
so I tested with that merge temporarily reverted.
@worldofpeace worldofpeace mentioned this pull request Sep 5, 2020
jtojnar and others added 2 commits Sep 6, 2020
This reverts commit 8425726.

This should have been reverted in #95358
but I forgot about it.
It's basically a partial revert of PR #95910.
I chose a temporar-ish solution, maximizing likelihood of fixing
this while minimizing rebuilds.  (20.09 process is being blocked)
@vcunat

This comment has been minimized.

Copy link
Member

@vcunat vcunat commented on 3106c30 Sep 6, 2020

Now cmake refuses to build on Darwin at all, breaking stdenv. This is the only cmake-related commit since it succeeded. Any idea? I would've reverted immediately if it weren't a huge rebuild on all platforms.

Quickly, please, nixos-unstable is blocked and progress towards 20.09 beta as well.

This comment has been minimized.

Copy link
Member Author

@sikmir sikmir replied Sep 6, 2020

fatal error: 'cmsys/Process.h' file not found looks strange, how it can be related to changes above...

This comment has been minimized.

Copy link
Member

@vcunat vcunat replied Sep 6, 2020

For reference here, the problem was somewhere else: #97296

@pbogdan
Copy link
Member

@pbogdan pbogdan commented Sep 6, 2020

I can't seem to be able to find it on Hydra but FWIW building stdenv shows that cmake-boot is also broken on Darwin.

I ran a bisect which pointed to #93560 but I don't really understand the changes (or know anything about Darwin) to know how to fix it (other than reverting those changes completely).

EDIT: oops, page didn't refresh so I didn't see vcunat's comment.
EDIT: I fixed texinfo by adding gettext on Darwin which seems to have the libintl stuff on that platform.

@FRidh
Copy link
Member Author

@FRidh FRidh commented Sep 6, 2020

I ran a bisect which pointed to #93560 but I don't really understand the changes (or know anything about Darwin) to know how to fix it (other than reverting those changes completely).

If reverting that change makes darwin build again, then that's good enough at this point. Please confirm you're able to build cmake(-boot) with that change reverted on darwin.

@pbogdan
Copy link
Member

@pbogdan pbogdan commented Sep 6, 2020

Yes, I reverted the change locally and cmake-boot now succeeds. I'm trying to build a full stdenv but it will take hours on my hardware (currently chewing through the stage3-stdenv stuff).

@pbogdan
Copy link
Member

@pbogdan pbogdan commented Sep 6, 2020

Can confirm Darwin stdenv builds with #93560 reverted, haven't built non-bootstrap cmake yet.

@pbogdan
Copy link
Member

@pbogdan pbogdan commented Sep 6, 2020

haven't built non-bootstrap cmake yet

It builds as well with the revert.

@FRidh
Copy link
Member Author

@FRidh FRidh commented Sep 6, 2020

Thanks for testing. Ugly work-around reverting the PR for Darwin #97296.

Temporary fix. We definitely do not want to have two separate
implementations.
@FRidh FRidh requested a review from Ericson2314 as a code owner Sep 6, 2020
@pbogdan
Copy link
Member

@pbogdan pbogdan commented Sep 6, 2020

@FRidh It looks like we may need to apply your to change to cc-wrapper as well at least (there is a fortran wrapper too I think but not sure how important that is), currently cmake-boot still won't build.
EDIT: patching cc-wrapper is sufficient to make cmake-boot build (didn't bother with any other wrappers).

@vcunat
Copy link
Member

@vcunat vcunat commented Sep 7, 2020

Haskell boostrap on aarch64 is broken. I can't see any relevant change (since master where it works), and I can reproduce the segfault on the shared box. EDIT: I opened #97407

Still, I don't think we should block merge to master because of this, in our situation.

@vcunat
vcunat approved these changes Sep 7, 2020
Copy link
Member

@vcunat vcunat left a comment

After a few thousand rebuilt darwin packages, I believe we're good enough to merge... at least in our dire situation.

@vcunat vcunat merged commit c1c85b9 into master Sep 7, 2020
2 of 4 checks passed
2 of 4 checks passed
tests tests
Details
action
Details
Wait for ofborg This failed status will be cleared when ofborg finishes eval.
Details
grahamcofborg-eval The branch this PR will merge in to does not cleanly evaluate, and so this PR cannot be checked.
Details
20.09 Blockers automation moved this from To Do to Done Sep 7, 2020
@vcunat
Copy link
Member

@vcunat vcunat commented Sep 7, 2020

The tested job succeeded on Hydra 🎉 Thanks to everyone who's helped; this staging iteration wasn't easy at all.

@acowley

This comment has been minimized.

Copy link
Contributor

@acowley acowley commented on 04fffd6 Sep 23, 2020

Do you think the issue I see with compilation failures is related to this? I had guessed that fixing the error I see would require a wrapper, but this commit message seems to be suggesting the opposite.

This comment has been minimized.

Copy link
Contributor

@acowley acowley replied Sep 24, 2020

I've opened a PR that seems to fix this for me, but I'm surprised nobody else has reported an issue. @antifuchs

@worldofpeace worldofpeace added this to In progress in 20.09 Blockers via automation Oct 5, 2020
@worldofpeace worldofpeace removed this from Done in 20.09 Blockers Oct 5, 2020
@worldofpeace worldofpeace moved this from In progress to Done in 20.09 Blockers Oct 5, 2020
@nixos-discourse
Copy link

@nixos-discourse nixos-discourse commented Oct 24, 2020

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/marketing-team-can-we-present-nix-nixos-better/6249/106

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
You can’t perform that action at this time.