On Tue, Sep 12, 2000 at 02:27:07AM -0400, Michael G Schwern wrote:
> On Tue, Sep 12, 2000 at 12:46:55AM -0500, J. David Blackstone wrote:
> > http://dev.perl.org/pm/pos.html mentions the following risk in perl6
> > development:
> >
> > > 4.We produce a slower interpreter.
> >
> > It is very easy to create something that is unacceptably slow and
> > not realize until toward the end of the process.
>
> True, true. A good idea to set up the benchmarks and run them periodically.
And it'll take a decent amount of time to build up a reasonably large
set of discrete benchmarks.
> My only concern is that they might touch of premature optimizations.
Possibly, but...
> It'll mostly be a matter of psychology.
...and communicating early and often that "premature optimization is the
root of all evil".
Write the benchmarks, then let the implementers to use them wisely.
It's not up to q/a to publicize performance numbers on a regular
basis until that feature freeze comes, when it's time to tweak perl6
to make it suitably faster than perl5.
Until that point, benchmarks serve as yet another way to test code
(and see progress).
Z.
