Something is broken with METIS, perhaps someone who is familiar with these packages can take a look. https://github.com/passagemath/upstream-scipoptsuite-deploy/actions/runs/16209293891/job/45766226347#step:4:107467

This would be a great addition!

@Joao-Dionisio I guess something has updated on the GitHub runner? Assuming @mkoeppe put the new CMAKE flags to debug the error and that they're NOT* the cause.

@mmghannam I don't have the capacity to look into this much right now. From the log file it looks like GKLib is being installed globally instead of locally? METIS is then not finding it as we're pointing to the wrong place. No clue why this has now broken.

Edit: @mkoeppe I imagine there's a very high chance this will get used for the upcoming SCIP 10 release

@Opt-Mucca Again, total newbie, but the only changes to the workflows have been not running tests on drafts, and [updating ubuntu to 22.04](https://github.com/scipopt/PySCIPOpt/pull/9769

The software installed on the GitHub runners is not relevant. The Linux builds run on the specified container images (manylinux), which are continuously updated. Relevant is the version of cmake that is on these images; my change involving the CONFIG_FLAGS is a workaround for making cmake accept the antiquated CMakeLists

@mkoeppe We've opened #27 (based on this PR). It's currently passing the pipelines, and looks like a successful linux-arm build! Thanks for the help! (Feel free to add any comments if you think something should be changed)