ENH: Implement subtraction for object-dtype Index #21981
Conversation
gfyoung
added
Enhancement
Indexing
|
|
||
| result = index - pd.Index([Decimal(1), Decimal(1)]) | ||
| tm.assert_index_equal(result, expected) | ||
|
|
There was a problem hiding this comment.
Can we also add tests where element-wise subtraction should fail.
There was a problem hiding this comment.
There are already some in the test just above this, but more couldn’t hurt. Will do.
There was a problem hiding this comment.
Ah, yes, thanks for pointing that out. Two things would be good there:
- We should rename that test above to be
test_sub_fail - Add a couple of tests where element-wise subtraction should fail
jorisvandenbossche
changed the title
You will need to give some more details, but let's do such a change in a separate PR anyway, so I would simply fix up and merge this one first. |
jorisvandenbossche
added
API Design
and removed
Indexing
pandas/tests/indexes/test_base.py
Outdated
| @@ -871,6 +871,18 @@ def test_sub(self): | |||
| pytest.raises(TypeError, lambda: index - index.tolist()) | |||
| pytest.raises(TypeError, lambda: index.tolist() - index) | |||
|
|
|||
| def test_sub_object(self): | |||
| # GH#19369 | |||
| from decimal import Decimal | |||
pandas/tests/indexes/test_base.py
Outdated
| index = pd.Index([Decimal(1), Decimal(2)]) | ||
| expected = pd.Index([Decimal(0), Decimal(1)]) | ||
|
|
||
| result = index - Decimal(1) |
There was a problem hiding this comment.
I don't think you are exercising rsub?
rsub is tested implicitly in the |
Codecov Report
@@ Coverage Diff @@
## master #21981 +/- ##
=========================================
Coverage ? 91.96%
=========================================
Files ? 166
Lines ? 50331
Branches ? 0
=========================================
Hits ? 46289
Misses ? 4042
Partials ? 0
Continue to review full report at Codecov.
|
|
@jbrockmendel thanks! |
git diff upstream/master -u -- "*.py" | flake8 --diff@jreback @jorisvandenbossche discussed briefly at the sprint. Merits more thorough testing, but I'd like to get the go-ahead to separate out arithmetic tests that are common to EA/Index/Series/Frame[1col] that are highly duplicative first.