Julian Rohrhuber

Just trying things out, this looks very good. If I understand correctly, @capital-G suggested that on the receiver side, the kwargs could be a dictionary instead of an array of...

I am able to crash sclang with any of the following: ```supercollider 3.performArgsAndKwArgs('+', nil) 3.performArgsAndKwArgs('+', []) 3.performArgsAndKwArgs('+', [], (aNumber: 8)) // these work as expected 3.performArgsAndKwArgs('+', [], []) 3.performArgsAndKwArgs('+', [],...

> When you use the array expansion `*[2, 3]`, for some reason the call goes through `Object:performList`. I don't know why? A possible reason is that you want to "reify"...

The tests are needed anyway, and for the binaries we build we can locally test. So why not?

So the point is that a method that forwards arguments should not have named arguments, but only `{ | ... args, kwargs| ... }`, right? I find it sematically somewhat...

> Also note, that the signature of performList must be a variable argument because that is what it is in the C++. Ah I see. All calls that had a...

Do you think you could make a list of all methods with ellipsis args that forward their arguments to a primitive? Just so we know how large this problem is...

Yes, this makes a lot of sense. Thanks for the claricfication of this really suprisingly complex issue! We wouldn't have expected that in the beginning, right? Now I am convinced...

The tests are running fine here, I'll poke around a bit to see if there are any boundary cases. So far this one is still crashing sclang (using the class...

Some git thing is messed up. I was on the wrong branch, it seems. But pull gives me a lot of merge conflics though I never touched any of these...