On Tue, Aug 21, 2001 at 12:59:52PM +0100, Andy Wardley wrote:
> I'm happy to put this into the core now if you like, but I'm thinking
> it might be better to keep it as a separate bundle for now.
...and of course, I reserve the right to change my mind at any
time. :-)
This is *REALLY* good stuff! Given that the Makefile.PL
make it quite clear that it's experimental, and it's totally
optional to build it, use it and/or have it as the default,
I'm thinking maybe it *should* go into the next core version?
After all, with these kind of benchmark results, it makes
sense for as many people to get the benefit as possible.
What do you think? Anyone else have any strong opinions
either way? Doesn't "Now more than twice as fast!" have
a nice ring to it? :-)
Check out these benchmarks results:
XS Module: $Id: Stash.xs,v 1.1 2001/08/20 20:04:49 dsteinwand Exp $
Template: ./testing.tmpl
Length: 2889 bytes
Status: XS output OK
======================================================================
Evaluating template with original stash 10 times...done.
Usr: 0.28s, Sys: 0.00s, Total: 0.28s, RSS: 0KB (0KB change)
Evaluating template with new XS stash 10 times...done.
Usr: 0.13s, Sys: 0.00s, Total: 0.13s, RSS: 0KB (0KB change)
Improvement: 2.15X
======================================================================
Evaluating template with original stash 100 times...done.
Usr: 2.77s, Sys: 0.00s, Total: 2.77s, RSS: 0KB (0KB change)
Evaluating template with new XS stash 100 times...done.
Usr: 1.31s, Sys: 0.00s, Total: 1.31s, RSS: 0KB (0KB change)
Improvement: 2.11X
======================================================================
Evaluating template with original stash 1000 times...done.
Usr: 27.80s, Sys: 0.05s, Total: 27.85s, RSS: 0KB (0KB change)
Evaluating template with new XS stash 1000 times...done.
Usr: 12.93s, Sys: 0.06s, Total: 12.99s, RSS: 0KB (0KB change)
Improvement: 2.14X
A
