Open h-vetinari opened 3 years ago
I think you need to add build:\n skip: true # [py<37]
for each output that uses python.
I think you need to add
build:\n skip: true # [py<37]
for each output that uses python.
I'm aware of that, but that's what shouldn't be necessary if there's a global skip on the top-level of the recipe. For now, I prefer to not litter the recipe with skips, and just leave the useless CI job hanging around (that just starts, skips everything & ends).
I think that is a Conda-Build issue. Can you please raise it there?
Hey @jakirkham, I can, but are we sure that it's a conda build
issue? After all, AFAIU, it's conda smithy
that generates the .ci_support
files (which is where the mistake happens).
conda-smithy performs re-rendering by having conda-build parse the recipe and run conda render (a feature of conda-build). So yes it is a conda-build issue
OK, thx
conda-smithy performs re-rendering by having conda-build parse the recipe and run conda render (a feature of conda-build). So yes it is a conda-build issue
Thinking more about it, I am no longer certain I am convinced... It could be that I misunderstood the behavior of conda render
. But if conda build
honors the skip (as we do observe), it means the outputs section in a skipped, rendered recipe should be empty. Can't conda-smithy check about this, and decide if a ci-support file is needed?
In https://github.com/conda-forge/modin-feedstock/pull/13, I updated
skip: true # [py<37]
, since upstream isn't compatible with 3.6 anymore.While trying to turn this recipe into a multi-output version, rerendering cause the 3.6 CI jobs to reappear (even though the build is skipped, so CI should stay green).