Pandas-dev

Download

pandas-dev@python.org

August 2025

1 participants
2 discussions

Next pandas community meeting - Wednesday January 22, UTC 18:00
by Joris Van den Bossche Jan. 14, 2026

Jan. 14, 2026

Hi all, A reminder that tomorrow we have the next bi-monthly (twice a month) dev call (Wednesday, January 22) at 18:00 UTC. Our calendar is at https://pandas.pydata.org/docs/dev/development/community.html to check your local time. The pandas Community Meeting is a regular sync meeting for the project's maintainers which is open to the community. All are welcome to attend! Video Call: https://numfocus-org.zoom.us/j/84024486573?pwd=QnVOUkozc3VzeXVsYnJEM1h4QVBG… Meeting notes: https://docs.google.com/document/u/1/d/1tGbTiYORHiSPgVMXawiweGJlBw5dOkVJLY-… Joris

1 2

ANN: pandas 2.3.2 - with preview of upcoming pandas 3.0 features
by Joris Van den Bossche Aug. 22, 2025

Aug. 22, 2025

Hi all, We are pleased to announce the release of pandas v2.3.2. See the release notes <https://pandas.pydata.org/docs/whatsnew/v2.3.2.html> for a list of all the changes. And as usual, the latest release can be installed from PyPI or conda-forge. While this patch release includes some bug fixes, we mostly want to draw attention to the opt-in features included in pandas 2.3, as a preview for changes coming in pandas 3.0. Starting with pandas 3.0, the dedicated string data type and the Copy-on-Write mode will be enabled by default. See Upcoming changes in pandas 3.0 <https://pandas.pydata.org/docs/whatsnew/v2.3.0.html#whatsnew-230-upcoming-c…> for more details. Summarizing here: - *String dtype*: Historically, pandas has represented string columns with NumPy's "object" data type. With pandas 3.0, the dedicated string data type ("str") will be inferred by default. This gives a much clearer user experience, and, if pyarrow is installed, will also give a performance boost for string operations. - *Copy-on-Write*: this is the technical term for the underlying implementation, but the main behaviour change is that the result of any indexing operation or any method returning a new DataFrame or Series, will always *behave as if* it were a copy. Thus, no longer modifying a DataFrame through chaining, and no longer a SettingWithCopyWarning! This are big changes, and changes that most likely will have an impact on your pandas code! Therefore, it would be great if you can help test those changes, and to make that easier they are included in pandas 2.3 behind a flag. You can enable both with: import pandas as pd pd.options.future.infer_string = True pd.options.mode.copy_on_write = True Check the string dtype migration guide <https://pandas.pydata.org/docs/user_guide/migration-3-strings.html> and Copy-on-Write user guide <https://pandas.pydata.org/docs/user_guide/copy_on_write.html> for more details about how to test, prepare and update your code to be compatible with those changes. Please report any issues you run into on the pandas issue tracker <https://github.com/pandas-dev/pandas/issues/new/choose>!

1 0