totally folding string operations

[Paolo Carlini]

Hi Giovanni,

very interesting and complex issues, thanks...

>... An 
>idea here would be to construct a new optimizer pass which transforms closed
>pairs of new/delete calls of constant size into alloca().
>
>Comments?
>
For now, only a very quick comment (I'm rather sure to say something 
sensible
at least about this :) I don't think this specific idea would work 
often, for
the simple reason that typically (by default in mainline, f.i.) we us pool
allocation and no actual new/delete pairs are involved for short string...

Paolo.