Test for pd.to_sql column error if data contains -np.inf

[arw2019]

• closes BUG: pandas.to_sql raises unexpected column error if data contains -numpy.inf #34431
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[arw2019]

Hey everyone - checking in in case there's anything I can do to keep this PR moving forward!

[mroeschke]

Looks like there is a merge conflict and there are some CI failures in the Travis build

[arw2019]

@mroeschke Ok, I've looked into that now.

It seems that the new test is failing:

=========================== short test summary info ============================
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemy::test_to_sql_with_negative_npinf[load_iris_data0-input0]
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemy::test_to_sql_with_negative_npinf[load_iris_data0-input1]
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemy::test_to_sql_with_negative_npinf[load_iris_data0-input2]
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemyConn::test_to_sql_with_negative_npinf[load_iris_data0-input0]
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemyConn::test_to_sql_with_negative_npinf[load_iris_data0-input1]
FAILED pandas/tests/io/test_sql.py::TestMySQLAlchemyConn::test_to_sql_with_negative_npinf[load_iris_data0-input2]
= 6 failed, 72578 passed, 1156 skipped, 1016 xfailed, 115 warnings in 1193.92s (0:19:53) =
The command "ci/run_tests.sh" exited with 1.

But when I run pytest -q pandas/tests/io/test_sql.py everything passes. Do by any chance you have an idea of what might be going on?

Either way, what's the best way to proceed on this PR? Should I try to fix the problem in this PR (happy to carry on but I'll need some help on where to start debugging - not sure what to try right now) or mark the test as xfail?

[mroeschke]

Looks like this issue isn't fully solved for MySQL based on the test failures.

If you would like to solve this issue for MySQL, that'd be great. It might need to be handled in sql.py.

Otherwise, we would also xfail the test for MySQL and leave the original issue open.

[arw2019]

Looks ready to go - happy to work on any improvements, though!

[mroeschke]

One comment otherwise LGTM

[mroeschke]

Thanks @arw2019. Nice PR as always