Issues installing bambi with pip

I’m not sure if this is only an issue with my current system, but when I try to install bambi with pip install bambi it fails after claiming that the dependency for numpy is numpy<1.22.0, >=1.16.1 which I know is wrong (i.e. related to an old version). However, the wheel that was downloaded is for bambi 0.9.1 which should depend on pymc and numpy>1.22. I tried running pip with --no-cache-dir but this made no difference. When I installed the version directly from github it installed correctly, but I still cannot fit a model because I get erros related to

ImportError: Version check of the existing lazylinker compiled file. Looking for version 0.212, but found None. Extra debug information: force_compile=False, _need_reload=True

which I’m guessing is related to some weird python dependency issue.

System details: manjaro linux 5.10.154-1 python 3.11.0 installed via pyenv I tried installing in a brand new virtualenv. I also tried specifying numpy version explicitly, as in pip install numpy=1.23.5 bambi=0.9.1 but this also didn’t work.

I’m not sure if this is a bambi packaging issue or something weird happening on my system, but any hints as to how to solve it would be greatly appreciated.

[omrihar]

Hi @tomicapretto I am indeed able to install bambi at the latest version, using pip install bambi. However you now have a different issue - since the pymc team decided to fork aesara and they no-longer rely on it, but you still do, if I don’t install it explicitly I cannot import bambi (get ModuleNotFoundError). I guess it would be an easy fix - just replace all instances of aesara with pytensor…

Thanks! I’ll close this one since it is indeed a different bug.

[tomicapretto]

@omrihar there’s a new release available. You should be able to install with pip now without this issue. Let me know if that works so we can close the issue 😃