switch with regex

[John Demme]

In my mind, a switch-case implies two things:  first, that one case will
match, and two that the compiler might be able to implement it faster than
if-else blocks.  Given:
uint i;
switch(i){
case 2: break;
case 5: break;
}

Only one case will be selected, so the compiler doesn't necessarily just
convert that block of code to the associated if-else if-else code-- there
may be a faster way to execute it.  The only way for the regex cases to be
compared is linearly, so the switch is just a nicer-looking if-else block.  

Also, multiple regex cases may match the string.  The behavior here is not
obviously defined.  Does it run all of the cases which match?  Only the
first one?

I think the behavior here would be too different and possibly inconsistent
from the current switch-case behavior to be worth it for syntactic sugar.

~John Demme

BCS wrote:

> Thoughts please:
> 
> pros?
> 
> cons?
> 
> why it won't work.
> 
> -------------
> switch(str)
> {
> case "[Tt]his"~~:
> writef("Doing this");
> break;
> 
> case "[Tt]hat"~~:
> writef("Doing this");
> break;
> 
> case ".*somthing.*"~~:
> writef("Doing somthing");
> break;
> 
> default:
> writef("Doing nothing");
> break;
> }