Update library requirements.txt#27
Conversation
|
Works successfully on windows 10 (python 3.10.11) |
There's still some errors being thrown which need to be addressed, but it runs!
|
@dahhei tried to install the venv on a mac and discovered shiboken needed to be updated from 5 -> 6 |
|
The app gets installed without any issue on mac M1. I can also train a classifier for single mouse dataset. However, when I try to export the classifier, I get the following error: Traceback (most recent call last):
File "/Users/chouda/tmp/jabs_app/JABS-behavior-classifier/src/ui/main_window.py", line 215, in _export_training_data
out_path = export_training_data(self._project,
File "/Users/chouda/tmp/jabs_app/JABS-behavior-classifier/src/project/export_training.py", line 92, in export_training_data
(1,), dtype=np.int)
File "/Users/chouda/tmp/jabs_app/JABS-behavior-classifier/jabs.venv/lib/python3.9/site-packages/numpy/__init__.py", line 313, in __getattr__
raise AttributeError(__former_attrs__[attr])
AttributeError: module 'numpy' has no attribute 'int'.
`np.int` was a deprecated alias for the builtin `int`. To avoid this error in existing code, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information. |
|
This branch would have particularly benefited from having more coverage on unittests.
Also I upgraded the singularity gui vm from buster -> bookworm, so should make the non-gui one also be bookworm (not bullseye). |
dahhei
left a comment
There was a problem hiding this comment.
The requirements.txt works without errors on my local M1 Macbook. Exporting a classifier also works fine on my end now.
|
A couple notes from testing/poking around:
|
|
Latest commit should clear the 2 issues above.
|
Conda environment was required for apple silicon but were reverted to reduce support complexity due to upgrade to pyside6 also resolving that. See #5 (comment) for more info.
Classifiers use new xgboost/scikitlearn Features use new numpy/shapely Software version always gets a bump
Updating the requirements.txt to not be 2-3 year old libraries.
There were a couple other changes to get compatibility working again.
Tests run at the beginning of this pull:
python3 -m unittest discover -s /behavior-classifier/tests/Ran 33 tests in 0.785s OKpython3 initialize_project.pyon a projectI don't have a bunch of dev environments to test, so these were all conducted inside singularity (running debian).
One of the fixed I needed to do (upgrade to shapely change) may address #5 (comment)
Also address #24
Would be good to have someone with a mac environment to share their experience.