id,node_id,number,title,user,state,locked,assignee,milestone,comments,created_at,updated_at,closed_at,author_association,active_lock_reason,draft,pull_request,body,reactions,performed_via_github_app,state_reason,repo,type
996352280,PR_kwDOAMm_X84rv1Fo,5794,Single matplotlib import,14371165,closed,0,,,7,2021-09-14T19:15:12Z,2022-08-12T09:06:30Z,2021-10-24T09:54:28Z,MEMBER,,0,pydata/xarray/pulls/5794,"Reduce number of imports inside functions. I think it helps making the code easier to read as well, as now you know that `plt` is always available.


- [x] Tests added
- [x] Passes `pre-commit run --all-files`
- [ ] User visible changes (including notable bug fixes) are documented in `whats-new.rst`

Seems to not be a major difference in initial imports from (my small sample of) repeated tests: 

This branch:
```python
%timeit -n1 -r1 import xarray

3.81 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.83 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.87 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.7 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.77 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.91 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.8 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)

np.mean([3.81, 3.83, 3.87, 3.7, 3.77, 3.91, 3.8])
Out[3]: 3.812857142857143
```

Main:
```python
%timeit -n1 -r1 import xarray

3.93 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.69 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.64 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.76 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.79 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.81 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)
3.68 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)

np.mean([3.93, 3.69, 3.64, 3.76, 3.79, 3.81, 3.68])
Out[4]: 3.7571428571428567
```","{""url"": ""https://api.github.com/repos/pydata/xarray/issues/5794/reactions"", ""total_count"": 2, ""+1"": 2, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,,13221727,pull
996475523,PR_kwDOAMm_X84rwPey,5796,Add asv benchmark jobs to CI,14371165,closed,0,,,16,2021-09-14T22:00:49Z,2022-08-12T09:01:15Z,2021-10-24T10:08:02Z,MEMBER,,0,pydata/xarray/pulls/5796,"Workflow based on the version from scikit-image. Modfied to have `asv.conf.json` inside a subdirectory and triggers every push if the PR has the `run-benchmark` label.

Notes:
* https://github.com/scikit-image/scikit-image doesn't have the same benchmark folder setup, for example config file is in root directory, other folder names.
* https://github.com/numpy/numpy has same folder name as sckit-image. config file is in the folder however.

References:
* https://labs.quansight.org/blog/2021/08/github-actions-benchmarks/
* https://github.com/scikit-image/scikit-image/pull/5424
* https://github.com/jaimergp/scikit-image/pull/1

Tests checked:

- [x] interp
- [x] pandas
- [x] repr
- [x] combine
- [x] datarray_missing
- [x] dataset_io - Skipped too difficult to understand. Some of the tests are possibly broken.
- [x] indexing
- [x] reindexing
- [x] rolling - Some division by 0 prints left.
- [x] unstacking

TODO:
* self.setup_cache
<!-- Feel free to remove check-list items aren't relevant to your change -->

- [x] Related to #4648
- [x] Tests added
- [x] Passes `pre-commit run --all-files`
- [x] User visible changes (including notable bug fixes) are documented in `whats-new.rst`

","{""url"": ""https://api.github.com/repos/pydata/xarray/issues/5796/reactions"", ""total_count"": 2, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 2, ""rocket"": 0, ""eyes"": 0}",,,13221727,pull