Update to python 3.11 #181
Conversation
netcdf4 to minimum of 1.6.4. Added flox as requirement to speed up resampling.
requires-python = ">=3.9, <3.12"
Nice, I have been using flox a lot recently. It has some really nice features (such as a groupby_bins that allows overlapping groups) |
|
Seems like there is an issue with string type variables in our data. My opinion is that xarray should be catching that before dispatching the writing to the netcdf4 library. Anyway, I would think that we can manually set the compression/encoding to |
|
Okay that worked. It is now stuck in to_mf_netcdf() and I can't seem to solve that one. |
If you make this change in datastore.py the test passes: |
|
Working on a patch, I'll push to this branch |
|
The problem was that on loaded dask arrays, strings are loaded as datatype objects and therefore were not identified as strings. As a result those datatype objects were also getting a zlib=True in their encodings. With this fix the other data_vars continue to be compressed |
|
Oh nice find! Dask arrays can be tricky that way |
CHANGELOG.rst
Outdated
| @@ -7,10 +7,13 @@ dev | |||
| New features | |||
|
|
|||
| * Improved the functionality of `merge_double_ended`, by adding a check that handles measurements missing in one channel while present in the other ([#171](https://github.com/dtscalibration/python-dts-calibration/pull/171)) | |||
| * Support for Python 3.11 | |||
| * Requiring netcdf4 >= 1.6.4 | |||
There was a problem hiding this comment.
Not really a new feature I'd think?
I think I like such a layout more for changelogs:
Fixed:
- Bug in abc
Added:
- Support for Python 3.11
Changed:
- Now requires
netcdf >= 1.6.4
netcdf4 to minimum of 1.6.4. Added flox as requirement to speed up resampling.