Closed marek-mihok closed 5 months ago
We probably want to sync publish conda step in release-wave with publish-conda job as well.
@mturoci what is it used for? Is it for cases when you want to publish conda without going through all release process? It has no runs yet.
Also is CondaDir: 'py/conda'
in publish-conda.yml
instead of CondaDir: 'py/h2o_wave/conda'
intentional?
And what exactly do you mean by "sync"?
Is it for cases when you want to publish conda without going through all release process?
Yes.
is CondaDir: 'py/conda' in publish-conda.yml instead of CondaDir: 'py/h2o_wave/conda' intentional?
Probably not. Seems like a leftover, but better check.
And what exactly do you mean by "sync"?
They should be the same. No matter which workflow I use, I want the output to be identical.
@mturoci synced. Can we try to run publish-conda
separately first to prevent release-wave
from potential failing?
The PR fulfills these requirements: (check all the apply)
main
branch.feat: Add a button #xxx
, where "xxx" is the issue number).Closes #xxx
, where "xxx" is the issue number.ui
folder, unit tests (make test
) still pass.The problem was that the package only supports the version of python it was built in. It can be solved by specifying
noarch: python
insidemeta.yaml
conda config file. The result is the single platform independent package supporting multiple python versions.Also update of the pipeline build action (
amauryval/publish_conda_package_action
) to version 2.0.3 is necessary, because it allows specifyingnoarch
value as thePlatforms
parameter.Closes #2227