I made a tool to manage Python packages inside Blender. It’s really bare bones currently. Let me know if you think this is a good idea. In Windows it’s best to open the terminal (Window > Toggle System Console) to see what’s happening when you press the buttons.
I made a new release (1.0.1). It should now print out the error message in the UI, as well as other messages generated by PIP.
I replicated your issue with simply using a non-existing module in the call, which means the addon can’t probably find PIP on your Blender installations Python. Also you can try turning “as local user” on/off if that has any effect.
Seems “ensurepip” only installs “pip3” module, not “pip”, I made changes to 1.0.2 that should use --default-pip which also installs “pip”.
Did you try Ensure PIP, Upgrade PIP and Install something (like for example “cffi”) first? The thing is probably write protected, but should work with user flag (in theory).
This is great!
Exactly what I needed. I am currently writing an add-on that requires several additional python packages (wheel, PyMCubes, scipy, pillow). My current workaround was to manually install all these packages from the command line to the local python installation that ships with Blender.
For details see here (section “Installation of local python libraries”):
It seems your package just makes it possible to install everything without leaving Blender. Works like a charm for me.
this is nice to have . i also suggest backing up a python directory and linking/mounting to that directory, so you can keep installed modules and changes accross blender versions
Unfortunately I can’t commit to supporting the add-on. If somebody else wants to do that, they have my blessing to fork the project and try to introduce it into the base Blender.
you dont need to , pip is bundled with blender as of v2.9 i think or 3.0 … aside from that i suggest people just use the template function of blender to integrate additional python libraries
Setup new libraries require some care, as you may break blender’s one with dependency.
Eg: if you setup latest scipy version, you’ll find yourself with a new numpy version too …
So be careful with that.
Sharing modules across blender versions is not an option as there are python version changes in the between - blender 2.8x / 2.9x / 3.x do not rely on the same python version.