BUG: Fix dtype inconsistency when appending to empty DataFrame #22703
Conversation
|
Hello @araraonline! Thanks for submitting the PR.
|
|
|
||
| df = DataFrame() | ||
| series = Series({'A': 1}, name='foobar', dtype='int64') | ||
| assert df.append(series).dtypes['A'] == 'int64' |
There was a problem hiding this comment.
pls construct the expected frame
and use assert_frame_equal
doc/source/whatsnew/v0.24.0.txt
Outdated
| @@ -775,6 +775,7 @@ Reshaping | |||
| - Bug in :meth:`DataFrame.replace` raises ``RecursionError`` when replacing empty lists (:issue:`22083`) | |||
| - Bug in :meth:`Series.replace` and meth:`DataFrame.replace` when dict is used as the ``to_replace`` value and one key in the dict is is another key's value, the results were inconsistent between using integer key and using string key (:issue:`20656`) | |||
| - Bug in :meth:`DataFrame.drop_duplicates` for empty ``DataFrame`` which incorrectly raises an error (:issue:`20516`) | |||
| - Bug in :meth:`DataFrame.append` would produce an inconsistent dtype when appending to empty DataFrame (:issue:`22621`) | |||
There was a problem hiding this comment.
use double backticla around DataFrame
|
This is not passing some tests that I did't notice before. Gotta see what's wrong |
gfyoung
added
Reshaping
|
Got a question. The problem ended up having a scope slightly bigger than what was mentioned in the Issue #22621 Now I'm working on a branch that fixes the dtype inconsistency when appending an array to a dframe, whether this dframe is empty or not. Should I create a new PR? |
|
@araraonline no you can just update this one |
|
I'm currently working on rewriting the DataFrame.append function. When it is ready, the issue related here will be resolved |
git diff upstream/master -u -- "*.py" | flake8 --diffedit: currently rewriting the DataFrame.append function