shlwapi: optimise path functions

[Andrew de Quincey]

On Monday 01 December 2003 20:33, Dimitrie O. Paun wrote:
> On December 1, 2003 03:26 pm, Alexandre Julliard wrote:
> > That's not possible, such a simple function cannot take 168us, unless
> > you have a 1Mhz CPU... How did you measure it?
>
> I think he measured it across the call:
>
>   start
>   CharNextW
>   end
>
> but it still doesn't add up. If his measurement is correct, we certainly
> have a high (it's an understatement, IIRC a syscall on Linux is on the
> order of a few us) call overhead. Maybe a bit of disassembly might help,
> but I can't imagine what could have gone wrong in such a trivial of a case.

I'm really distrustful of those results; something very weird seemed to happen 
during that test. I will re-run it again and let you know.