Poetry VS Pixi when building Python CLI apps

[T145]

When using Poetry, I can have something like this in my pyproject.toml:

[tool.poetry.scripts]
myproj = "myproj.main:app"

Where app is a CLI function in main, and after installing be able to run myproj.
Using Pixi's task, to my knowledge this is the equivalent:

[tool.pixi.tasks]
myproj = "python src/myproj/main.py"

And it still requires pixi run myproj. Is there a way to accomplish what Poetry does with Pixi?

[synapticarbors]

My understanding is that [tool.poetry.scripts] is a python build tool way of creating an entry point. You can do the same using other build backends (e.g hatch. See also the entry point spec).

It seems it would be best practice to use one of the build backends to install your local project, here contained in src, into a pixi env, and then it would just be available as myproj on the command line once you enter pixi shell or in a task. This can either be done with a pypi-dependencies that specifies the local project (see: https://github.com/prefix-dev/pixi/blob/main/examples/pypi-source-deps/pixi.toml#L29-L30), or via a task that uses pip or uv to install it, like in https://github.com/prefix-dev/pixi/blob/main/examples/docker/pyproject.toml#L46-L48

[T145]

Ok so the first example is interesting! I have a Python project with only pyproject.toml containing the project settings. When creating a Python CLI project, do I then need to create an initial Pixi project with pixi init, then do pixi init myproj --pyproject?