PERF: no need to check for DataFrame in pandas.core.computation.expressions #40445
Conversation
|
cc @jreback you might be more familiar with the history behind this. I suppose at some point, DataFrames were passed directly to |
jreback
added
the
Algos
| _array2 = _frame2["A"].values.copy() | ||
|
|
||
| _array_mixed = _mixed["D"].values.copy() | ||
| _array_mixed2 = _mixed2["D"].values.copy() |
There was a problem hiding this comment.
we really need to use fixtures for this module.....can you create an issue
There was a problem hiding this comment.
Yeah, opened #40497 for this
| @@ -170,16 +173,9 @@ def testit(): | |||
|
|
|||
| op = getattr(operator, opname) | |||
|
|
|||
| result = expr._can_use_numexpr(op, op_str, left, left, "evaluate") | |||
There was a problem hiding this comment.
so this will just raise now? if so can you add a test
There was a problem hiding this comment.
Actually, it seems we just return False. So I could leave that as a test, but note that this is a file-internal helper function (leading underscore) that's within pandas never called with a DataFrame. So I would basically be testing something that is not expected to have any specific behaviour.
|
@jreback OK here? |
|
thanks @jorisvandenbossche |
In practice, this case is only covered by specific tests using DataFrame in
test_expressions.py, not from actual usage through ops (so I just updated the failing tests to not use any DataFrame as input to those checks).Although our test suite might not be the best suited to check that, since many tests are small and thus don't use the numexpr path... (fixing that in #40463)