Re: explicitly defining table length + _ARG proposal

[Florian Weimer <fw@...>]

* Mark Hamburg:

> A smart optimizer could recognize the following and optimize them to avoid the likely n-squared costs:
>
> 	select( '#', ... )
> 	select( i, ... )

And

  #{...}
  ({...})[i]

could be optimized at compile time, and they would do the right thing
with explicit table lengths. 8-)

(Perhaps it's even possible to make {...}[i] parse as expected?)