Package 'tensorflow-macos' is listed as a dependency of itself

[marygiotandoi]

I am building the local environment of a Python project, on a MacBook Pro (Apple M1 Max). However, I cannot install anymore tensorflow-recommenders = "~=0.7.3" using poetry.

I'm getting the following error:

$> poetry lock --no-cache                                                                                                    
Creating virtualenv <path to venv>
Updating dependencies
Resolving dependencies... (4.4s)

Package 'tensorflow-macos' is listed as a dependency of itself.

I'm using the following versions:

• Python 3.11.1
• Poetry: 1.8.2
• Pip: 24.0

To reproduce this, create an empty directory, make sure you have Python and poetry setup, paste the following as your initial pyproject.toml

[tool.poetry]
package-mode = false
name = "project-name"
description = "Project description."

[tool.poetry.dependencies]
python = "~=3.11.0"
tensorflow-recommenders = "~=0.7.3"

And then run

poetry lock

[elisasimoni]

I solve it by downgrading tensorflow-recommenders to 0.7.2, works like a charm, the problem is a recursive dependency in the new package

[celha]

I'm facing the same error. Is there any other way to solve it? I want to use something other than version 0.7.3.

[marygiotandoi]

I solve it by downgrading tensorflow-recommenders to 0.7.2, works like a charm, the problem is a recursive dependency in the new package

Thanks! The trick I am using right now is to lock without upgrading un-tached dependencies in the .toml file (poetry lock --no-update). I can do this because at some point I did manage to create a .lock file with tensorflow-recommenders = "~=0.7.3". The problem only started to arise with the latest releases of tensorflow in March 2024. I hope it gets solved.