Fix skip handling

[jakirkham]

It appears that commit ( 75dc6bf ) from PR ( #479 ) results in some unintended skips as demonstrated by PR ( conda-forge/viridis-feedstock#4 ). Will need to investigate a fix that still has the intended behavior.

Edit: Another example is PR ( conda-forge/qcodes-feedstock#1 ).

[jakirkham]

cc @johanneskoester @jni

[jakirkham]

When inspecting the meta, we find that all cases result in skip being True. We are going to need look for some more specific information if about the skip conditions, but will need to figure out where.

[johanneskoester]

Alternatively, one could still try to find the conda/conda-build combination where the computation works without a problem. This should be doable by using my test case, deactivating the skipping, and checking where it crashes.

[jakirkham]

Thanks for the suggestion, @johanneskoester. I'll keep it in mind when I have time to do more debugging.

[jakirkham]

Fixed by PR ( #493 ).

[jakirkham]

The fix is being released in conda-smithy 2.2.2.

xref: conda-forge/conda-smithy-feedstock#73