OK - I’ve started making addons which are 90% my nodegroups plus a little python code made with the help of ChatGPT (a can hear you’re cries of dis-approval)
But I’m finding every time we update the code we have to
Try VScode with the Blender development extension. A little cumbersome to set up but once it works, it’s a breeze. Plus it comes with a nifty auto-load script that gets rid of having to registering your classes in the right order. Otherwise you may set up symlinks between your dev environment and the blender addon folder, that’s pretty much how the VScode extension works. Also, please don’t use notepad ^^ unless you meant notepad++.
I just wish I knew a way to do the same with PyCharm here - my C drive is to close to full to try to install VSCode correctly so I’m still doing to the smaller version add-ons and then once proofed, bringing the parts into the main addon and testing that.
That is not a helper. It may look that it makes things go, but it deepens you into ‘dark territories’ very quickly!!
You could use a versioning tool (i.e. git), and just create a patch with the differences between different versions.
Of course, AI code can change substantial between requests for modifications (which an human programmer avoid it at all costs), and your patches could be the size of the whole addon! You just need to apply the patch over the current system, and reload…
If the addon identification is the same, you just need to reload the addons…
you could add a call for this anywhere in Blender. (AI won’t ever do this for you!)