API: Stop modifying SQL column and names, and warn when pertinent. #6902
Conversation
|
Looking good, I think this does reflect consensus! Maybe also add a test to check that a column name with a space is written correctly? And what warning messages do you get in nose? |
| @@ -769,6 +765,11 @@ def _create_sql_schema(self, frame, table_name): | |||
| } | |||
|
|
|||
|
|
|||
| _SAFE_NAMES_WARNING = ("The spaces in these column names will not be changed." | |||
There was a problem hiding this comment.
A space after changed. is missing
|
Edited in response to @jorisvandenbossche's review and added a test for a bug in the original PR: DataFrame column names must be converted to strings. |
|
Ah, good, the integer column name error was still a to-fix on the #6292 list. Checked it as fixed! Looking good for the rest. Ready to merge for you? |
|
nosetests still spews a straw warning. Know a trick for finding out which test is to blame? I am pretty sure |
|
*straw -> stray |
|
which warning do you get? If you just execute the file (tests_sql.py, there is a main that calls nose), then I see the progress of the tests in my terminal, and you can see for which test it is generated. |
|
BTW, there is still a warning that you should not care about for the moment (about not being a SQLAlchemy engine, at the moment you get that always when using the legacy, so just the first test using a DBAPI connection is triggering this). |
|
Oh OK. I thought that warning was due to me. Then this is ready to merge. |
API: Stop modifying SQL column and names, and warn when pertinent.
|
No it was not you, but I think we should do something about the warning (but that is a discussion for legacy-status issue). Merged, thanks! |
Closes #6796
If any columns names written
to_sqlvia a legacy connection would have been modified by previous versions of pandas, a warning is issued, both on table creation and row insertion. But no names are modified.There is a stray warning message in nosetests. Otherwise good, I think. Does this reflect our consensus?