@salkin-mada Thanks for this! I will look over it during the weekend.

@salkin-mada Hi again, sorry for the late response. I tried out the PR, and it's definitely nice to have more snippets for subclasses!

We do get some additional snippets though that produces some noise (imho), for instance we get *newFromDesc for a lot of subclasses that inherits from MultiOutUGen since there is no way in SuperCollider to specify the scope of methods (private/public) and that particular method is used internally by a lot of *ar/*kr creation methods.

Just to be clear, this was also the case before this PR, only now we get even more :) I'm starting to think that we might need a more granular approach to creating snippets? Maybe by ignoring some methods such as *initClass, and maybe also some of the Abstract* classes, since they shouldn't be instantiated directly in most cases. I'll have to think some more about this some more, but we can continue to discuss this in this PR for now.

@davidgranstrom Indeed there is a lot of additional noise, which is really not ideal.
To rethink the approach of how we do this - I am in favor of.
To be able to get all the subclasses and methods we want "for user convenience" is something I would really get excited about.
Lets discuss further and share ideas on how this could be done.
For now I could try to hardcode some exclusions for the most "aggressive" parents and clear some room for the inheritors and the things we really want. Maybe just start with the Ugen subsubclasses i.e. the noise classes etc. But making hardcoded exceptions is def not the way to go in the long run.
Could we somehow begin to formulate exakli what we want?

also see #63, a new parser for the classes and methods with more flexibility would be a thing to consider

@salkin-mada Yes, I agree. I think a good start would be to study https://github.com/supercollider/supercollider/blob/develop/SCClassLibrary/scide_scqt/ScIDE.sc in detail and see what/if we could reuse any of the methods.

I'm currently working on better function signature hints ("argument hints") and I think I will need something more robust for that as well than what we have now.

Closing this for now since it probably will be resolved by using better introspection from https://github.com/davidgranstrom/scnvim/tree/topic/arghints-lua

The plan is to generate snippets etc. directly from lua based on the introspection data generated by sclang.