Closed kendal-s closed 2 years ago
You could turn the example you gave in the PR body into a test without the DataFrames dependency by using plain ol' vectors, e.g.
x = [TimeSpan(0, 100)]
x .= TimeSpan(0, 200)
@test x == [TimeSpan(0, 200)]
Thanks Kendal! If someone is willing, it might be nice to backport this for a v0.2.9 release since (AFAIK) a lot of stuff doesn't yet support TimeSpans v0.3.
Thanks for the support on this! I'm not familiar with the process for backporting but I'd love to get insight on how to do it
Sure thing! Basically it just involves some minor git gymnastics. Starting from an updated clone of the repository,
git checkout release-0.2
git checkout -b kks/backport-pr-43
git cherry-pick -x -e 902c49a210ead85584d5d21373d64f1a1163b88f
The cherry-pick will likely cause a conflict since it changes the Project.toml in a way that doesn't match the state on the release-0.2
branch. The resolution would be to bump the patch version according to the current v0.2 version. Once the change is made and the conflict is resolved,
git add -u
git cherry-pick --continue
git push origin kks/backport-pr-43
then open a PR against the release-0.2
branch. The base branch for a PR is configurable in the GitHub UI. Once that's all squared away, feel free to request me for a review. 🙂
Currently broadcasting TimeSpans returns an error without wrapping the TimeSpan in a
Ref
. It would be nice to be able to broadcast a Timespan without having to wrap it!Example: