This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Document eval vs. sub in Benchmark
authorHugo van der Sanden <hv@crypt.compulink.co.uk>
Tue, 1 Apr 1997 22:31:55 +0000 (23:31 +0100)
committerChip Salzenberg <chip@atlantic.net>
Tue, 1 Apr 1997 00:01:35 +0000 (12:01 +1200)
commit446534a1e286d6fde0fd79424865d3cf1291546b
treee2d9973dae347624625080db5c5d189b38aea4e0
parent7e5dee47ee5bc43e90e14d8c27c33a42dd0a26ae
Document eval vs. sub in Benchmark

Subject: Re: Patch for Benchmark.pm

In <9704011547.AA29906@toad.ig.co.uk>, Tim Bunce writes:
:Simple: calling functions in perl is expensive.
Yeah, I simply hadn't expected it to be *that* expensive.

:> - ? "sub { package $pack; my(\$_i)=$n; while (\$_i--){&\$c;} }"
:> + ? "sub { package $pack; my(\$_i)=$n; while (\$_i--){&\$c();} }"
:It's cheaper to _not_ add () so I'd recommend leaving it off.
I considered that, but felt it was more correct not to pass through the
parameters that were received. My new patch ducks it, by shifting @_
out before it gets there.

I also looked at having two caches for empty string/empty sub timings,
but was unable to get satisfactory results. I'll come back to that one,
but I don't think it is urgent for 5.004.

This patch is for the Benchmark.pm in _96: it repeats stuff from my
previous patch, but avoids stuff that was obsoleted by other changes
in _96.

p5p-msgid: 199704012231.XAA00225@crypt.compulink.co.uk
lib/Benchmark.pm