How to use custom python dependencies

[RealTachyon]

This isn't a bug, but rather a question for how to properly use this extension.

I'm planning to write an addon that makes use of third party python libraries, as well as libraries that I've written myself.

Usually my working strategy is to create a new VSCode workspace with a pyproject.toml and then use poetry to set up a virtual environment.
In the pyproject.toml I would then set up dependencies like

[tool.poetry.dependencies]
attrs = "*"
rich = "*"
colorama = "*"

my_utils = { path = "../my_utils", develop = true }

The first three in this example would be third party dependencies.
The last one is a custom python library that lives in a sibling folder. For this one, it would be important to have it installed in development mode so that changes to it are recognized and used in higher level programs.

After installing the virtual environment with poetry, those dependencies would live in .venv\Lib\site-packages.

But this approach doesn't seem to be enough for this extension. How can I ensure that my addon has access to the required dependencies?

If possible, I would like to avoid installing the individual dependencies by hand in my Blender folder, since requirements can change, and I would like to keep the installation dynamic.

[Mateusz-Grzelinski]

For starters, have you seen https://docs.blender.org/manual/en/dev/advanced/extensions/addons.html#bundle-dependencies

In this add-on we download pip, and then install deps to blender modules path using --target option in pip. That is not ideal.

The most proper way is to bundle deps with your addon, see the link

[RealTachyon]

Thank you for the reply and the link!

But I'm still not sure how I should approach the bundling of my dependencies.
I can certainly get behind keeping them all local, bundled with my addon. But I'm not sure how to actually achieve this yet, especially with the requirement that some of the dependencies need to be installed in develop mode.

In this add-on we download pip, and then install deps to blender modules path using --target option in pip. That is not ideal.
When you say "add-on" do you mean the blender_vscode extension for VSCode? Or an actual addon for blender?
And in either case, what is the recommended pipeline I should follow if I want to both bundle my dependencies and still make use of blender_vscode?

I'm just generally still trying to figure out a good pipeline for development, so pointers there would be very appreciated!

[MYSELFABHISHEKK]

When making a Blender add-on with Python dependencies, it's best to bundle everything inside your add-on folder. You can do this by installing your packages using pip with the --target option, like this: pip install requests --target=./my_addon/modules. This keeps all the dependencies local and avoids touching Blender’s internal Python setup.

To make sure Blender can find these packages, you just need to add the modules folder to sys.path in your __init__.py. This tells Blender where to look when importing your dependencies.

If you're developing a custom package alongside your add-on, you can symlink it into the modules folder. That way, any changes you make will reflect immediately without reinstalling. On Windows, if symlinks don’t work, you can just copy the folder instead.

To speed up development, use the blender_vscode extension for VSCode. It lets you run and debug Blender from VSCode, which makes testing your add-on much easier. Just make sure it's pointing to your Blender executable and you're good to go.

In short: install dependencies to a modules folder, add that folder to sys.path, use symlinks for editable packages, and use blender_vscode for a smoother development experience.I was going through the same problem and this worked for me ..