--count

[Ovid]

--- On Sat, 9/8/08, Andy Armstrong <andy at hexten.net> wrote:

> >  I notice as a side effect that the
> --nocount shaves 3 or  
> > 4 wallclock seconds on the Test::Harness test suite :)
> 
> Wow. That sounds like an area for further investigation.
> I'm surprised  
> terminal I/O is so costly.

I'm not sure I'd be overly concerned.  Yes, we want TH to run as fast as possible, but I've noticed that for larger test suites, the harness speed is almost irrelevant to the suite performance.  I once ran 16000 OK tests in about 2 seconds, but our test suite, for the same number of tests, took over a half hour.

> Any debate we have about that will tend to go round in
> circles...

I'm coming around to your point of view.
 
Cheers,
Ovid