DOC: Remove tips and tricks section page from the wiki

[datapythonista]

This page: https://github.com/pandas-dev/pandas/wiki/Tips-&-Tricks-for-pandas-dev is mostly outdated, or provides general Python information.

Personally, I'd just keep the last section on PANDAS_TESTING_MODE. Not sure if the best would be to move it to the CI part of the docs: https://pandas.io/docs/development/contributing.html#testing-with-continuous-integration

Or I'd probably just add a comment where that parameter is being used: https://github.com/pandas-dev/pandas/blob/master/.travis.yml#L46

Anything else that people would like to move to the docs from that page?

[saloni30agr]

Hey @datapythonista,
I would like to work on this issue. From the above description, it looks like, it would be better to move last part - Enabling Deprecations when running tests - to https://github.com/pandas-dev/pandas/blob/master/.travis.yml#L46. Should I go ahead with this?

[datapythonista]

I think that make sense, and having a PR will facilitate the discussion. But can't promise we'll move forward without further discussion.

[keshavgu]

@datapythonista Is the problem still active?

[datapythonista]

@datapythonista Is the problem still active?

Thanks for the triaging. In my opinion we could delete the wiki page now. @simonjayhawkins do you think that wiki page can be deleted, or do you think some content is still useful, and should be moved to the docs?

[simonjayhawkins]

Some observations on the sections in Tips & Tricks for pandas dev

---

Testing for versions

This refers to the use of LooseVersion. Quick search does not find LooseVersion in the docs.
The example uses LooseVersion(np.__version__)>='1.7.0'. For numpy we use _np_version_under1p16 etc from pandas.compat.numpy
For testing we generally use skip_if_no etc from pandas.util._test_decorators unless we need to version for reasons other than skipping tests.
We have a convention for the imports where we always import pandas.util._test_decorators as td and then use the decorator as, for example, @td.skip_if_no("pyarrow", min_version="0.15.0").

I think we should capture these conventions in our contributing style guide, imports and testing sections.

---

Skipping an entire file full of tests on a failed import of a module named a_module

We no longer use this pattern. We mainly use, for instance, @td.skip_if_no_scipy as a test function decorator or test class decorator.
With pytest, we can skip at module level. With just a brief search, I didn’t find any module level skips in the code. I think our preference is to retain test classes for namespacing and to use the class decorator.

This section also refers to including a message so that when testing we can see why things were skipped.

I think we should ensure these conventions are captured in our contributing style guide, testing section.

---

Nose

We now use pytest, but the operations mentioned can also be done with pytest.

Exclude a particular test (e.g., a test that's failing you don't care about)

$ pytest pandas/tests -k ‘’not some_test_name’

Fail fast (on first failure - useful with git bisect)

$ pytest pandas/tests -x

Run with ids (and just re-run tests that failed)

$ pytest pandas/tests --lf

Print out skipped tests and their messages

$ pytest pandas/tests -rs

Personally I feel this information is out-of-scope for the pandas documentation.

---

Misc

Reading from a url

The convention is to use urlopen from pandas.io.common. This is not currently documented but is not used much in the code.

Reading a file named file.txt that's inside of a zip file named file.zip

The convention is to use ZipFile from pandas.io.common. Again this is not documented elsewhere.

Enabling Deprecations when running tests

Addressed by #31008

I think we should ensure the conventions for urlopen and ZipFile are captured in our contributing style guide, testing section

---

PRs addressing any of the above welcome.

[vijaysaimutyala]

@simonjayhawkins @datapythonista I would like to work on the above. This is my first time contributing here. As per above comments, I suppose parts of the wiki should be going here.

https://pandas.io/docs/development/code_style.html

Please correct me if I'm wrong.

[simonjayhawkins]

As per above comments, I suppose parts of the wiki should be going here.

https://pandas.io/docs/development/code_style.html

That's where I think we should now be adding new content.

see also #33851 for current discussion/investigation about improving the code style docs. @Stockfoot @moaraccounts

[vijaysaimutyala]

Thanks @simonjayhawkins. I was reading your comments on keeping the first PR's as minimal as possible and submitted my first PR that covers just urlopen and ZipFile in the style guide with #34366. Hope it looks ok.

I've a question on this. I was using VS Code for editing the rst file. I noticed VS Code has the reStructuredText extension that enables to live preview the rst file. However I see the below error while trying to start the live preview.

I've used the conf.py that's present in the root directory of the project. Sphinx does seem to have installed well since I was able to build the docs. Do you have any suggestions on this ?

[MarcoGorelli]

Do you have any suggestions on this ?

I wasn't able to get that working either, I usually just build the page I'm interested in seeing and open the html in the browser. e.g.

cd doc
python make.py clean
python make.py --single development/code_style.rst

[egd001]

Hi,

Is this issue closed or can I help?

Regards

[simonjayhawkins]

Thanks @egd001 the Testing for versions and Skipping an entire file full of tests on a failed import of a module named a_module sections from #30964 (comment) are outstanding

[debnathshoham]

Hi, I am new to open source contributions.
Is this issue open? It's still not clear to me, where to move the tips and tricks section to?

[datapythonista]

Looks like all the useful content has already been moved. Only left is how to compare versions in Python with setuptools (not worth having in the pandas docs IMO, more for a stackoverflow question). And a section on nose, which I'm not aware of being used anymore.

If there are no objections I'll delete the wiki page soon.

[datapythonista]

Deleted the page.