[
https://issues.apache.org/jira/browse/BEAM-9064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17010307#comment-17010307
]
Chad Dombrova commented on BEAM-9064:
-------------------------------------
Thoughts (which I will repeat in the mailing list discussion):
I've gotten mypy completely passing on my own branch. It was a lot of work and
the longer it lingers the more work it becomes. So let's focus our efforts on
getting that merged and integrated into the lint jobs _first._ From there we
will learn a few things: 1) How difficult it is for end users to keep the
tests passing (i.e. the time and frustration levels with mypy alone), and 2)
how much more work it will be to also get pytype passing (i.e. how divergent
the two tools are). Based on that feedback we can decide whether we want a
pytype failure to _prevent_ a PR from merging, or merely be informative
(failures could be something that the google team tracks on the side through
metrics/analytics).
tl;dr If both tools are actually quite similar, then it shouldn't be much
more of a burden for users to maintain both. But if they're not, or users are
already struggling with mypy on its own, then we should hold off introducing
pytype as a requirement.
> Add pytype to lint checks
> -------------------------
>
> Key: BEAM-9064
> URL: https://issues.apache.org/jira/browse/BEAM-9064
> Project: Beam
> Issue Type: Improvement
> Components: sdk-py-core, testing
> Reporter: Udi Meiri
> Assignee: Udi Meiri
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> [~chadrik]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
