API / CoW: constructing DataFrame from DataFrame creates lazy copy #50499
Conversation
# Conflicts: # pandas/core/generic.py
# Conflicts: # pandas/core/generic.py
| @@ -951,6 +951,10 @@ Indexing | |||
| - Bug in :class:`BusinessHour` would cause creation of :class:`DatetimeIndex` to fail when no opening hour was included in the index (:issue:`49835`) | |||
| - | |||
|
|
|||
| Copy on write | |||
| ^^^^^^^^^^^^^ | |||
| - Bug in :class:`DataFrame` constructor not tracking reference if called with another :class:`DataFrame` (:issue:`50499`) | |||
There was a problem hiding this comment.
For the equivalent Series change I listed this in the "Copy-on-Write improvements" section instead of seeing it as a bug fix (it was a known gap left out of the initial implementation, and it also changes behaviour, so at this stage I wouldn't yet label it as a bug fix)
| @@ -205,6 +205,7 @@ | |||
| to_arrays, | |||
| treat_as_nested, | |||
| ) | |||
| from pandas.core.internals.managers import using_copy_on_write | |||
There was a problem hiding this comment.
This import can now be removed (after a merge with main)
| and new_data is self._mgr | ||
| or not copy | ||
| and using_copy_on_write() | ||
| ): |
There was a problem hiding this comment.
Can you explain a bit how this change is related? We are passing here a Manager object to the constructor, I suppose, and you only changed the constructor when it receives a DataFrame?
There was a problem hiding this comment.
Ah sorry, this is unrelated to the constructor change, I'll make a separate pr.
Currently, when reindexing with copy=False we don't track references, causing problems.
There was a problem hiding this comment.
I also have some skips in #50536 related to reindex, there are several things not yet fully working it seems
doc/source/whatsnew/vX.X.X.rstfile if fixing a bug or adding a new feature.cc @jorisvandenbossche