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
sundials: 5.3.0 -> 5.6.1 #107286
sundials: 5.3.0 -> 5.6.1 #107286
Conversation
Result of 3 packages failed to build:
1 package built:
|
nixpkgs-review on x86_64-linux runs without errors. |
cc @flokli per bmcage/odes#107 (comment) / LLNL/sundials#19 ? |
8d4fac5
to
1f52d9e
Compare
Ouch, I don't have the hardware to properly test this on darwin. It might be this regressed upstream (not sure if there's any darwin testing on their side). @rb2k, interested to take a look? |
As is, there's some other problems when running nixpkgs-review pr 107286 from my side. I think the last force-push might have been a bit dicey:
That being said, once it works I would run "nm -gU" on the binary to see if the library is mentioned there. |
1f52d9e
to
fbf3fbd
Compare
Correct. I removed the I removed the patch from our expression although that PR was closed because upstream wrote that it was included in v5.5.0. I didn't find the cmake variables mentioned in the PR in the main Could it be we should set |
Result of 1 package marked as broken and skipped:
4 packages failed to build:
4 packages built:
|
I am not sure how to test this. I tried
but got
I tried
but got
|
Thanks for the report @rb2k. Interesting Also, I'd like to know if adding a cmake flag |
That's weird and seems unrelated to me - you just spawn a
That makes sense - you can't run commands that create files in directories in
|
Yes probably something I missed when upgrading to Big Sur
Ok the examples build but I get segfaults - I am guessing the linker is picking up some incompatible versions of the libs - I continue to investigate. |
Ok I removed the brew versions and now I get
I changed this to
and now everything builds but I get runtime errors such as
|
@idontgetoutmuch please don't build sundials manually via make, or via I'm not sure about the state of sandboxing on darwin - you might want to remove any globally installed libraries first. |
I think I found out why upstream has introduced this regression: This line: is different than this: Should I open a PR upstream? |
Never mind, I'll open the PR upstream. Justified by: https://github.com/jtojnar/cmake-snips#assuming-cmake_install_dir-is-relative-path |
fbf3fbd
to
cf99e7f
Compare
@idontgetoutmuch if you could please retest now, I haven't opened the PR upstream but I updated the PR here to include the patch I'll send to them at: https://github.com/LLNL/sundials/compare/master...doronbehar:rpath_install_dir?expand=1 |
Tested - works like a charm :) |
Update some cmake flags and be more explicit regarding index size.
24213a2
to
442f1ec
Compare
Great, Updated the PR with an updated comment linking to the PR upstream. |
If someone can run a final nixpkgs-review on darwin it'd be great 🙏. |
running it :) |
Result of 1 package marked as broken and skipped:
1 package failed to build:
7 packages built:
|
Thanks. The fact that octaveFull fails to build while octave isn't is not a regression of this PR: https://hydra.nixos.org/job/nixpkgs/nixpkgs-20.09-darwin/octaveFull.x86_64-darwin So I merge this. |
Update some cmake flags and be more explicit regarding index size.
Motivation for this change
Update.
Things done
Tested
octave
to build with this update.sandbox
innix.conf
on non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
./result/bin/
)nix path-info -S
before and after)