Re: extended rx.el

[Stefan Monnier]

> By the way, rx tends to insert more shy groups than necessary.  Is
> that likely to have a significant performance impact on the regexp
> engine in typical circumstances?

Normally no.  A shy-group turns into a nop, so it's very efficient.
But it can prevent some optimizations that recognize special cases.
E.g. IIRC \(?:.\)*\n will be (measurably) slower than .*\n because
the optimizer is too stupid to notice it's the same thing (if the
optimization was applied after byte-compilation it would work, tho).


        Stefan