Prettify function tostring

[Denneisk]

Outputs function(ARGS): RET for tostring operations including printTable.

[Denneisk]

(Don't merge yet will also put E2Descriptions for functions since they weren't included)

[Vurv78]

Thanks, forgot to update the tostring post making functions store their arguments and return types

[Vurv78]

Suggested change

	return "function(" .. self.arg_sig .. ")" .. (self.ret and ": " .. self.ret or "")
	return "function(" .. self.arg_sig .. ")" .. (self.ret and (": " .. self.ret)) or ""

[Denneisk]

This makes the LuaLS upset

[Vurv78]

Need another set of parens wrapping or "" I assume

[Denneisk]

Actually, should functions be allowed in arrays?

[Vurv78]

Set the default function to one that throws this error. This is adding overhead to every function call

[Denneisk]

I wanted to do that but then it gives an error for bad parameters instead, which I thought would be unideal.

let BadFunc = table()[1, function]

BadFunc(1) # Error: Expected void

If you think that's worth sacrificing, then that's fine. The error stems from the same root I guess.

[Denneisk]

As a compromise, maybe the error could be thrown after if f.arg_sig ~= sig? Then it would only be adding that overhead to errors. Compare it to a nofunction constant.

[Denneisk]

Forgot error throws an uncatchable error so I hope this is okay (it's just a convenience function for E2Lib.Debug.Error.New, not a method that uses self in a meaningful manner).

[Vurv78]

Ok. I'm at my PC. Can we please not make completely different changes to a PR after already getting an accepting review (basically making it pointless?) Either leave it as draft or make another PR.

[Vurv78]

At this point I'm just leaning toward not having a default value for function. It's pointless trying to accommodate old E2. Ideally @strict would do this but chances are that'd break 99% of strict chips. So seems it's time for that new versioning directive.

[Denneisk]

Right, my bad.