If I install this, is there a way to revert the pytorch env to its original form?

[stephantual]

The node is flag in the manager as (and I quote) "Due to changes in the torch package and versions of many other packages, it may disrupt your installation environment. "
Let's imagine i install this in prod and it accidently switches pytorch to a version i can no longer use with other nodes, will i be able to revert pytorch (and other changes) back to original by uninstalling through the manager or are we looking at something more complex?
Thank you.

[NeedsMoar]

If you're on Windows you won't be able to start comfy, it installs / installed a non-CUDA and older version of torch via pip which causes a failure to start due to lack of CUDA support. It's likely it'll do the same thing if you're using directml or any other backend that needs its own version (isn't in the basic torch package). Unless they've updated it.

You would then need to uninstall torch, torchvision, and friends and reinstall the cuda / directml version before you could access comfy again. Oh and delete the custom_node directory for this so it doesn't accidentally have the requirements run again and do the same thing.

[stephantual]

understood, thank you - good to know, too!! I appreciate your help.