@GrahamcOfBorg build python27Packages.flask-babel python37Packages.flask-babel python38Packages.flask-babel

Did you actually run nix-review? I tried to verify, but the build of python38.pkgs.flask_sqlalchemy failed because of this update.

Did you actually run nix-review? I tried to verify, but the build of python38.pkgs.flask_sqlalchemy failed because of this update.

I was a little premature in checking that. I'm running it right now locally in a VM and it's going terribly slow. Will post fixes/output as appropriate.

Result of running nixpkgs-review pr 83314:

https://github.com/NixOS/nixpkgs/pull/83314
6 package failed to build:
apache-airflow python37Packages.flask-appbuilder python38Packages.apache-airflow python38Packages.flask-appbuilder sage sageWithDoc

6 package built:
python27Packages.flask-babel python27Packages.ihatemoney python37Packages.flask-babel python37Packages.ihatemoney python38Packages.flask-babel python38Packages.ihatemoney

• flask-appbuilder is fixed in follow-up pythonPackages.flask-appbuilder: 2.1.6 -> 2.3.0 #83315
• apache-airflow is in pretty bad shape due to pinning too many requirements. See my previous comments and attempts to fix dr-pr-colorlog fix branch
• Looking into sage now. Might just be an unpin fix, TBD. Long build :\

Did you actually run nix-review? I tried to verify, but the build of python38.pkgs.flask_sqlalchemy failed because of this update.

@timokau I didn't see flask_sqlalchemy in the dependencies when running nixpkgs-review pr 83314 (see above).

Yes, I don't see it anymore either. Maybe it was fixed by some other change in master, or maybe I somehow got it confused with flask-appbuilder.

I'm the maintainer of sage, if you upload the log as a github gist or something I might be able to help.

Edit: If you want to look into it yourself, note that you can run individual tests with

nix-build -E 'with (import ./. {}); sage.tests.override { files = ["src/sage/path/to/some/test"]; }'

Which speeds up debugging a lot.

Hm. Might be some spurious timeout because nix-review was building so many things at once. That shouldn't happen, but maybe it did. Would love to see the log of the original failure. If its still there, you might be able to dig it up with

find /nix/var/log/nix/drvs/ -name '*-sage-tests*' -mtime 2

which will find all logs of the sage testsuite from builds that occurred in the last two days. You can then inspect the actual logs with nix log while stripping one slash, i.e. translate

/nix/var/log/nix/drvs/82/sxc5m1a5rn82i8rd5pmr0f2bbqssyh-sage-tests-9.0.drv.bz2

to

nix log 82sxc5m1a5rn82i8rd5pmr0f2bbqssyh-sage-tests-9.0.drv

@GrahamcOfBorg build sage

I looked for the sage-test deriv using your commands, and couldn't find anything. This was the closest I could find. Looks like tests passed. Not sure why build failed. Looks like the ofborg build passes too.
https://pastebin.com/1zkub9Ee

Hm, very weird. Either way its likely not this PR's fault.

Regarding flask-appbuilder, isn't there a way to order the commits to avoid the intermediate failure?

should just be an update, as the newest version of appbuilder uses flask-bable 1.0.0 https://github.com/dpgaspar/Flask-AppBuilder/blob/7b796071bae32c5bac30f1d2f2b29b42814b7709/setup.py#L52

@jonringer I fixed it in #83315. I think these commits basically need to be made in lockstep (i.e. same PR?) for them to not fail build. I can merge the two PRs if that'd work?

I see, I probably would have done it as part of this PR, but thats fine

Thanks for the tip. I'll keep that in mind in future.