BUG/TST: reset setitem_copy on object enlargement

[jreback]

related #5581
closes #5597

[jreback]

@jtratner, cc @TomAugspurger

this fixes the warning in tests/test_index.py. I don't believe this changes any thing else.

can you give a whirl.

[TomAugspurger]

Yep that did it.

[jreback]

@TomAugspurger pls give this updated one a whirl... can you post a script of what you are doing...I can add as tests...

[TomAugspurger]

I'm just running the index.py test.

On master:

(pandas-dev) ~/E/p/l/p/s/p/p/tests (master↑ ) pwd
/Users/tom/Envs/pandas-dev/lib/python2.7/site-packages/pandas/pandas/tests
(pandas-dev) ~/E/p/l/p/s/p/p/tests (master↑ ) nosetests test_index.py
................................................................................................................................................................./Users/tom/Envs/pandas-dev/lib/python2.7/site-packages/pandas/pandas/core/generic.py:1029: SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_index,col_indexer] = value instead
  warnings.warn(t, SettingWithCopyWarning)
.................
----------------------------------------------------------------------
Ran 178 tests in 1.131s

OK

This PR:

(pandas-dev) ~/E/p/l/p/s/p/p/tests (jreback-copy_warn=) nosetests test_index.py
..................................................................................................................................................................................
----------------------------------------------------------------------
Ran 178 tests in 0.928s

OK

The causing the warning was this one:

    def test_set_value_keeps_names(self):
        # motivating example from #3742
        lev1 = ['hans', 'hans', 'hans', 'grethe', 'grethe', 'grethe']
        lev2 = ['1', '2', '3'] * 2
        idx = pd.MultiIndex.from_arrays(
            [lev1, lev2],
            names=['Name', 'Number'])
        df = pd.DataFrame(
            np.random.randn(6, 4),
            columns=['one', 'two', 'three', 'four'],
            index=idx)
        df = df.sortlevel()
        self.assertEqual(df.index.names, ('Name', 'Number'))
        df = df.set_value(('grethe', '4'), 'one', 99.34)  # <<<< this line
        self.assertEqual(df.index.names, ('Name', 'Number'))

[jtratner]

Why don't we default SettingWithCopy to error in the test suite? (and then
make sure the default is warning and test what happens with warning/None)

[jreback]

@TomAugspurger ok...that was my test case as well

@jtratner in theory you could do that after this PR. In test/test_indexing.py, the options are set (for testing), but other than that the default ('warn') is used.

Let me change it

[jreback]

@jtratner I can set it in a couple of the likely test suits...any way to change it 'globally' though? (only for testing)?

[jreback]

the right way to do this is to have a pandas unittest class then have all tests inherit from this, and have the setup in a sub always call this, but more work that what I am doing

[jreback]

too hacky?

I don't want to just import the test suite to actually trigger this so I think have to actually call it at the top of each file (because you could just call that file independently). We don't have a 'global' test suite setup ATM

[jtratner]

Hmm, that's true.

[jtratner]

Why not just define a class with a setUp method and have other modules
inherit that?

[jreback]

@jtratner you could do that

but then you have to make sure that an existing setUp calls it and change the inheriting from unittest every where

that would then allow say ignoring of FutureWarning and such in an easy way (but so will this)

[jtratner]

Okay, however you want to do it is fine with me. Maybe there's a global
nose setup that you can do via environment variable?
On Nov 25, 2013 8:54 AM, "jreback" [email protected] wrote:

@jtratner https://github.com/jtratner you could do that

but then you have to make sure that an existing setUp calls it and change
the inheriting from unittest every where

that would then allow say ignoring of FutureWarning and such in an easy
way (but so will this)

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/5584#issuecomment-29218769
.

[jreback]

looks like you can do setUpClass a class method (assuming you sub-class).... then wouldn't need to have the setUp actually call super...

[jreback]

@jtratner alright...revised to baseclass all TestCases....whoosh!

[jtratner]

Okay I'll take a look tonight.

[jreback]

@jtratner give this a look see pls

[jreback]

@jtratner ok with this?

[jtratner]

Why did this get deleted?

[jreback]

was all commented out - afaik never used

[jtratner]

@jreback this looks fine, though I realized we could have just defined a single setup_package and teardown_package in each of the testing modules instead:

So could have just done this:

def setup_package():
    pd.set_option('chained_assignment','raise')
    #print("setting up: {0}".format(cls))

def teardown_package():
    #print("tearing down up: {0}".format(cls))
    pass

and then in test __init__.py files just do

from pandas.util.testing import setup_package, teardown_package

so knowledge for the future.

[jreback]

ok that's prob about the same amount of code I guess
good 2 know though