We require contributors to sign our Contributor License Agreement and we don't have one on file for @tkelman.

In order for us to review and merge your code, please e-sign the Contributor License Agreement PDF. We then need to manually verify your signature, merge the PR (conda/infrastructure#1076), and ping the bot to refresh the PR.

pre-commit.ci autofix

CodSpeed Performance Report

Merging #5565 will not alter performance

_{Comparing tkelman:dynamic-tags (26eb40e) with main (9c253dc)}

Summary

✅ 5 untouched benchmarks

The one serial test failure seems like an unrelated fluke? Not sure why the parallel failures are having issues installing newer py-lief, maybe because of the repodata patch (conda-forge/conda-forge-repodata-patches-feedstock#926)?

@conda-bot check

Not sure why the parallel failures are having issues installing newer py-lief,

I think those recipes are pulling from defaults, which only has 0.12.3 so far. I think you might need to do something like getattr(base, "attr", fallback) to include both options and drop the >=0.15 constraint.

Other than that, LGTM! I'll approve once tests are passing.

Ah, and we need a news/ file too, sorry!

those recipes are pulling from defaults

Ah good call, I'm used to only working over in conda-forge

might need to do something like getattr(base, "attr", fallback) to include both options and drop the >=0.15 constraint.

I think this should do the job. Would be nice if there were a more concise way to write out the tests to be compatible with both versions but I can't think of one offhand.

The last PR that put in the fallbacks for py-lief 0.14 API changes #5228 had some temporary commits during the PR to run CI tests with the new py-lief version from a rc label, let me know if you think we should go through the same steps here. I don't want to accidentally break the tests when 0.15 gets pulled into defaults

and we need a news/ file too

Done, used the one from #5228 as a template

I'm going to put lief=0.16 on the conda-forge's main label conda-forge/lief-feedstock#51

With this PR, it seems to succeed for me locally.

Did you run the conda-build tests locally or do you mean it was working to locally build some things from conda-forge? Either way thanks for checking. Looks like we need a maintainer to approve the github actions workflow so CI runs again.

Whenever this gets released conda-forge should be able to drop the upper bound on the lief dependency.

Did you run the conda-build tests locally or do you mean it was working to locally build some things from conda-forge?

I ran the tests locally; I hoped that this would suffice.

Thanks to whoever authorized the CI run. Got a bunch of timeouts, http 500 errors, and a few mismatched checksums that all seem unrelated - would probably go away on a re-run. Is there some bigger degraded infrastructure issue today on github actions or conda servers?