I played around with various ways of speeding up the Perl function. This was the fastest:
sub get_date_perl2 {
$_[0] =~ m/[|](\d{4})(\d{4})[|]/;
return $2.$1;
}
I added this to the benchmarks as Perl2:
Benchmark: timing 50 iterations of Inline, Ovidian, Perl, Perl2...
Inline: 19 wallclock secs (18.72 usr + 0.01 sys = 18.73 CPU) @ 2
+.67/s
Ovidian: 17 wallclock secs (17.08 usr + 0.08 sys = 17.16 CPU) @ 2
+.91/s
Perl: 25 wallclock secs (24.48 usr + 0.08 sys = 24.56 CPU) @ 2
+.04/s
Perl2: 24 wallclock secs (23.36 usr + 0.14 sys = 23.50 CPU) @ 2
+.13/s
Rate Perl Perl2 Inline Ovidian
Perl 2.04/s -- -4% -24% -30%
Perl2 2.13/s 5% -- -20% -27%
Inline 2.67/s 31% 25% -- -8%
Ovidian 2.91/s 43% 37% 9% --
However, I wouldn't put this in production code for a 5% saving. I would probably use
split and
unpack for clarity.
--
John.