DOC: Not running clipboard examples in the doc

[datapythonista]

• xref CI/DOC: Building the documentation with azure-pipelines #26648, CI/DOC: fix clipboard on travis #26103
• tests added / passed
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

The clipboard example in the docs is failing when building in azure (#26648), and had to be previously fixed in travis (#26103). I don't think there is any need to run those examples, and we can keep things simpler. This adds the same value to the doc reader, and doesn't require to set up a clipboard service in the doc builds. Also, we're thinking on having a doc build for windows (#26574) not sure if that could cause new problems.

CC: @jorisvandenbossche

[jorisvandenbossche]

Yes, this is a good idea:-)

But, I think we should still be careful about the actual underlying issues though. The fact that this was not working for the docs, might be an indication that we are actually not running any tests for clipboard on azure.

[jorisvandenbossche]

How is this related to #26685 ?

[datapythonista]

Sorry, it was supposed to be #26103, github autocomplete does weird things when you copy/paste issues.

Agree on the tests, the skipping when a dependency is missing is quite dangerous. Do you want to open an issue for it? Or I can do it.

[jorisvandenbossche]

There is #26428 for the general discussion (not specific to the clipboard tests, but triggered by it).
But we should probably open a specific issue for clipboard+Azure.

[codecov]

Codecov Report

Merging #26700 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #26700      +/-   ##
==========================================
- Coverage   91.88%   91.87%   -0.01%     
==========================================
  Files         174      174              
  Lines       50701    50701              
==========================================
- Hits        46588    46584       -4     
- Misses       4113     4117       +4

Flag	Coverage Δ
#multiple	90.41% <ø> (ø)	⬆️
#single	41.9% <ø> (-0.12%)	⬇️

Impacted Files	Coverage Δ
pandas/io/gbq.py	78.94% <0%> (-10.53%)	⬇️
pandas/core/frame.py	97% <0%> (-0.12%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 649ad5c...96d2d9d. Read the comment docs.

[codecov]

Codecov Report

Merging #26700 into master will decrease coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #26700      +/-   ##
==========================================
- Coverage   91.88%   91.87%   -0.01%     
==========================================
  Files         174      174              
  Lines       50701    50701              
==========================================
- Hits        46588    46584       -4     
- Misses       4113     4117       +4

Flag	Coverage Δ
#multiple	90.41% <ø> (ø)	⬆️
#single	41.9% <ø> (-0.12%)	⬇️

Impacted Files	Coverage Δ
pandas/io/gbq.py	78.94% <0%> (-10.53%)	⬇️
pandas/core/frame.py	97% <0%> (-0.12%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 649ad5c...96d2d9d. Read the comment docs.

[WillAyd]

lgtm

[jreback]

I am still wary that these tests are not running, why not have the docs fail if these are not building? rather than paper this over?

[datapythonista]

I agree that we can be skipping those tests unintentionally, but I don't think that the doc warnings is how we should know. And the job and dependencies for the docs is not the same as for the tests, so we can be correctly executing the .to_clipboard() method in the docs, but still not running the tests.

I think the importorskip can already skip unwanted tests silently (like if a library is installed but fails to import), but I guess it's the best option. But I don't think we should have the condition in https://github.com/pandas-dev/pandas/blob/master/pandas/tests/io/test_clipboard.py#L15 but instead be specific on when we want to run those tests (if we want to run only for specific builds, may be something like python --skip-clipboard?)

[jreback]

@datapythonista that's not the question that I had. We have not been testing the clipboard tests for quite some time AND DIDN't KNOW this. (because the config was not correct) The docs are IMHO a last resort at least someone notices.

[datapythonista]

Let's close this then.

But as I said, the docs build and the tests builds are independent. Seeing the warning in the docs don't mean that the tests are skipped, and not seeing it doesn't mean that they are being run. I think what I propose would make sure that we run them when we want, but I guess I'm not understanding you.

[TomAugspurger]

I would recommend dedicated pytest calls on CI jobs where we know the clipboard tests should be run. I imagine we can use some pytest wizardry to select just the clipboard tests and fail if they are skipped.

…

On Jun 8, 2019, at 18:07, Marc Garcia ***@***.***> wrote: Closed #26700. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.