You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, in today's meeting the consensus was that such testing needs to happend out-of-band from the normal PR presubmit testing. I have therefore reverted the already-merged PRs which had only that change:
Does --pre in the constraints file do what we want? I'd be more comfortable with that approach if we also had a line to log the version like we have in bigquery. e.g. session.run("python", "-c", "import grpc; print(grpc.__version__)")
Also:
There are some packages like pandas and pyarrow which publish their pre-releases to a separate index. These need to be installed separately.
We usually specify a top bound to our dependencies, but pre-release versions might be outside those bounds. pip will happily install a non-prerelease version rather than the latest pre-release version to be within those bounds. (We worked around this by using --no-deps in the bigquery session.
/cc @busunkim96, @tswast, @crwilcox
I had originally worked to add
--pre
to thetesting/constraints-3.9.txt
files of the "manual" libraries:However, in today's meeting the consensus was that such testing needs to happend out-of-band from the normal PR presubmit testing. I have therefore reverted the already-merged PRs which had only that change:
For the already-merged PRs which contained more changes, I have created PRs which back out only the constraint change:
For the not-yet-merged PRs, I have backed out the constraint change, and re-titled them to signal the remaining changes:
Going forward, we need to work out how to do this testing out-of-band.
prerelease_deps
session tonoxfile.py
(not run by default, becausenoxfile.py
lists explicit default sessions That PR sets up a separatepresubmit
job to run that specificnox
session.continuous
) builds?The text was updated successfully, but these errors were encountered: