html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/pull/5690#issuecomment-896515958,https://api.github.com/repos/pydata/xarray/issues/5690,896515958,IC_kwDOAMm_X841b792,41898282,2021-08-11T05:39:18Z,2021-08-19T17:55:49Z,CONTRIBUTOR,"## Unit Test Results
         6 files  ±0           6 suites  ±0   53m 7s :stopwatch: ±0s
16 218 tests ±0  14 482 :heavy_check_mark: ±0  1 735 :zzz: ±0  1 :x: ±0 
90 504 runs  ±0  82 323 :heavy_check_mark: ±0  8 175 :zzz: ±0  6 :x: ±0 

For more details on these failures, see [this check](https://github.com/pydata/xarray/runs/3374642776).

Results for commit 378b9020. ± Comparison against base commit 378b9020.

:recycle: This comment has been updated with latest results.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,965939431
https://github.com/pydata/xarray/pull/5690#issuecomment-898142259,https://api.github.com/repos/pydata/xarray/issues/5690,898142259,IC_kwDOAMm_X841iJAz,5635139,2021-08-13T02:51:59Z,2021-08-13T02:51:59Z,MEMBER,"> > I would suggest that new tests use `-> None:` so mypy checks them, and we build up some tests for our typing through that — any thoughts?
> 
> I'm not opposed to suggesting `-> None` for tests, but I would hesitate to require it:
> 
>     * It's not unusual for tests to be incompatible with type checking, e.g., because they are checking invalid input, or rely on tricks like mocks.
> 
>     * It's also a fair amount of boilerplate to add to code that already has a lot of boilerplate (for things like parametrized tests & fixtures), so readability could suffer.

Suggesting is fine. 

I might vote to strongly suggest — it does worry me that a bunch of our type annotations are wrong and we don't have a mechanism to assess that. 

But for sure if someone is using mocks etc, then no need. And I'm not suggesting everything (e.g. fixtures) needs to be annotated — adding `-> None:` will let mypy test the types it already knows about.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,965939431
https://github.com/pydata/xarray/pull/5690#issuecomment-897396180,https://api.github.com/repos/pydata/xarray/issues/5690,897396180,IC_kwDOAMm_X841fS3U,1217238,2021-08-12T06:57:56Z,2021-08-12T06:57:56Z,MEMBER,"> I would suggest that new tests use `-> None:` so mypy checks them, and we build up some tests for our typing through that — any thoughts?

I'm not opposed to suggesting `-> None` for tests, but I would hesitate to require it:
- It's not unusual for tests to be incompatible with type checking, e.g., because they are checking invalid input, or rely on tricks like mocks.
- It's also a fair amount of boilerplate to add to code that already has a lot of boilerplate (for things like parametrized tests & fixtures), so readability could suffer.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,965939431
https://github.com/pydata/xarray/pull/5690#issuecomment-897280353,https://api.github.com/repos/pydata/xarray/issues/5690,897280353,IC_kwDOAMm_X841e2lh,5635139,2021-08-12T01:41:28Z,2021-08-12T01:41:28Z,MEMBER,"Unfortunately most test files fail mypy when their tests are changed to `-> None` — which enables mypy. This is mostly because they reuse variables for different types, though on occasion because our typing isn't quite correct.

I would suggest that new tests use `-> None:` so mypy checks them, and we build up some tests for our typing through that — any thoughts?","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,965939431