Make -NaN and -nan default NA values

[Bklyn]

closes #5952

[dsm054]

Probably test_default_na_values should be brought into alignment?

[jreback]

@Bklyn yep on test_default_na_values...in theory that should fail with your change (until its fixed)....does it?

[Bklyn]

Sorry for the noob-ishness, but what is the suggested way to run the tests against not-yet-installed version of pandas (e.g. from my working copy)? I'm getting import errors if I run the test script by hand unless I "setup.py install" first.

At any rate, the tests do not fail if I don't change them, presumably because they're not trying the new -nan values. Will fix the code and update the PR with the tests for completeness.

[jreback]

@Bklyn make this part of the same test (which SHOULD fail if the _NA_VALUES are not updated)

e.g.

assert_array_equal(_NA_VALUES,pandas.io.parsers._NA_VALUES)

This way the test have to be updated as well

[jreback]

@Bklyn

python setup.py build

then

nosetests pandas/io/tests/test_parser.py

[Bklyn]

Is there anything left for me to do on this PR?

[jreback]

pls add a release notes entry (put in 0.13.1 API),

rebase and squash to 1 commit: https://github.com/pydata/pandas/wiki/Using-Git

[jreback]

update this too (actually add to the text): http://pandas.pydata.org/pandas-docs/dev/io.html#na-values

[Bklyn]

OK, I think thats got it.

[jreback]

just add this line also to doc/source/release.rst under API changes as well. we use the whatsnew to 'announce' important changes, and the release.rst as a log of them

[Bklyn]

Should be there now.

[jreback]

@y-p @jtratner ok by you?

[ghost]

fine by me.

[jreback]

thanks @Bklyn

nice little PR, and you were very responsive!!

[Bklyn]

Thanks for the hand-holding and the feedback. You guys (and perhaps gals?) do a great job with pandas and your attention to code and documentation quality is impressive.

[jreback]

welcome! feel free to tackle any open issues!

[ghost]

Yeah, we need a gal commiter.

Well, a ninja that happens to be a gal.