Upgrade pandas dependency to current version #933
Assignees
Labels
Comments
forman
added
enhancement
pont-us
added a commit
that referenced
this issue
Mar 1, 2024
pont-us
added a commit
that referenced
this issue
Mar 1, 2024
Partially addresses Issue #933. test_timecoord.py generated some expected values from date-time string literals in a non-standard format, which pandas 1.x parsed correctly but pandas 2.x parses incorrectly. I have changed these literals to ISO 8601 format and they now parse reliably.
pont-us
added a commit
that referenced
this issue
Mar 1, 2024
Some long-deprecated unittest assertion aliases were removed in Python 3.12. This commit replaces usages of those aliases with the preferred method names: assertEquals -> assertEqual assertRegexpMatches -> assertRegex Partially addresses #933 (since removing the pandas <2 version pin means that newly created xcube environments get Python 3.12 instead of Python 3.11 by default).
pont-us
added a commit
that referenced
this issue
Mar 1, 2024
- Change two incorrectly parsed date literals to ISO 8601 - Use 'D' rather than (no longer supported) 'd' to specify days as a time unit. Partially addresses Issue #933.
pont-us
added a commit
that referenced
this issue
Mar 1, 2024
Python 3.12 has removed the wstr_length field from its internal PyCompactUnicodeObject struct, changing the size of string objects, which broke some tests in test_cache.py due a hard-coded expected size. This commit make the test calculate the expected size at runtime instead. Related to Issue #933.
4 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The latest version of xcube 1.4.0. constrains for its environment the pandas version to pandas >=1.3,<2. The latest version of pandas is, however, 2.2.1. As a consequence, an installation into a typical, up-to-date Python data science environment leads to a downgrade of one of its core packages to an outdated major version. Such downgrade may cause source code developed with pandas >2.0 to fail, which is likely unacceptable for most use cases and thus prevent the usage of xcube altogether.
To avoid the above-described downgrade, xcube source code should be adapted to run also with pandas >2.0. There is no alternative solution as compatibility with the latest pandas version is of critical importance for the user uptake of xcube.
This feature is requested by a user who want wants to use xcube in the Jupyter Hub of the Copernicus Data Space Ecosystem (CDSE, https://documentation.dataspace.copernicus.eu/Applications/JupyterHub.html) where pandas is currently installed in version 2.1.4.
The text was updated successfully, but these errors were encountered: