Implement Path.__deepcopy__ avoiding infinite recursion #30198
Conversation
fa7ed8b to
68d25b3
Compare
|
I tried the same strategy with EDIT: this is actually easier to fix by just doing the shallow copy without |
8258353 to
8fbb889
Compare
|
I changed this to remove the metaclass and implement the |
8fbb889 to
aa94dda
Compare
lib/matplotlib/path.py
Outdated
| if memo is None: | ||
| memo = {} |
There was a problem hiding this comment.
This is not needed, memo is always a dict.
The default value for memo is also not needed. The memo argument is always passed to __deepcopy__().
There was a problem hiding this comment.
Yes, when called via copy.deepcopy, but this __deepcopy__ method is also aliased to deepcopy (just below the GitHub patch context) that is part of the public API of this class
There was a problem hiding this comment.
we could change our deepcopy to call copy.deepcopy internally (which might bet better?)
2d93b8e to
a76ab5a
Compare
To deep copy an object without calling deepcopy on the object itself, create a new object of the correct class and iterate calling deepcopy on its __dict__. Closes matplotlib#29157 without relying on private CPython methods. Does not fix the other issue with TransformNode.__copy__. Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: Elliott Sales de Andrade <quantum.analyst@gmail.com>
without calling copy.copy
db1a305 to
5c7c915
Compare
…finite recursion
|
Thank you @jkseppan ! This is a better solution than my "reach up and touch private methods" version. |
…198-on-v3.10.x Backport PR #30198 on branch v3.10.x (Implement Path.__deepcopy__ avoiding infinite recursion)
Implement
Path.__deepcopy__avoiding infinite recursionTo deep copy an object without calling deepcopy on the object itself, create a new object of the correct class and iterate calling deepcopy on its
__dict__.Closes #29157 without relying on private CPython methods.
In a separate commit, fix the other issue with
TransformNode.__copy__.PR summary
PR checklist