JuliaGaussianProcesses / KernelFunctions.jl

Julia package for kernel functions for machine learning
https://juliagaussianprocesses.github.io/KernelFunctions.jl/stable/
MIT License
267 stars 32 forks source link

Remove chainrule and test for `SqMahalanobis` #539

Closed simsurace closed 9 months ago

simsurace commented 9 months ago

Just some cleanup work to remove remnants of the Mahalanobis kernel which has been absent for a while.

codecov[bot] commented 9 months ago

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Comparison is base (ec19a94) 90.43% compared to head (a8d07c5) 90.34%.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## master #539 +/- ## ========================================== - Coverage 90.43% 90.34% -0.09% ========================================== Files 52 52 Lines 1390 1378 -12 ========================================== - Hits 1257 1245 -12 Misses 133 133 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

simsurace commented 9 months ago

Hi @willtebbutt, the version was already bumped, but I haven't released it just yet (was hoping to figure out the failures on master first). Do we merge this anyway and release it with the other fixes once they are done?

willtebbutt commented 9 months ago

Ah okay, I hadn't realised we were in this situation. As a rule, we never leave an unreleased version of the package on master -- if a PR (with patch bump) is good enough for a release.

Could you please release the current version of master, then bump + release this? I grant you that it's annoying that there are failures on master, but unless you think that they're related to a PR that has recently been merged, I don't feel that they ought to disrupt the release cycle.

simsurace commented 9 months ago

No the remaining failures are likely due to relying on some internals in Julia, and were introduced by the Julia 1.8 -> 1.9 update. I will do as you suggest.