_____________________________________________________
Jeff japhy Pinyan: Perl, regex, and perl hacker.
s++=END;++y(;-P)}y js++=;shajsj<++y(p-q)}?print:??;

japhy, that's what I would've thought too, but I timed it using Benchmark and the times were actually comparable. One runthrough tr would be a second or two faster, the next one s/// was faster by a second or two. However, as the iteration count went up, the substitution was actually faster consistently.

use Benchmark;

timethese(250000, {
    one => sub {
        $string = "blah+blah+blah";
        $string =~ tr/+/ /;
    },
    two => sub {
        $string = "blah+blah+blah";
        $string =~ s/\+/ /g;
    }
});

print "\n";

timethese(2000000, {
    one => sub {
        $string = "blah+blah+blah";
        $string =~ tr/+/ /;
    },
    two => sub {
        $string = "blah+blah+blah";
        $string =~ s/\+/ /g;
    }
});
[download]

NOTE: I did use counts between 250,000 and 2,000,000 as well.

Maybe its my system, but I thought that this was strange.

Amel - f.k.a. - kel

Which version of perl are you using?

5.005_03 seems to have a really slow tr///, that actually is slower than s/// (singe celeron 466)

Benchmark: timing 250000 iterations of one, two...
       one:  1 wallclock secs ( 0.83 usr +  0.00 sys =  0.83 CPU)
       two:  1 wallclock secs ( 0.74 usr +  0.00 sys =  0.74 CPU)

Benchmark: timing 2000000 iterations of one, two...
       one:  7 wallclock secs ( 6.71 usr +  0.00 sys =  6.71 CPU)
       two:  6 wallclock secs ( 5.93 usr +  0.00 sys =  5.93 CPU)
[download]


However, 5.6.1 (on a dual p3 800) tr seems to be almost twice as fast.

Benchmark: timing 250000 iterations of one, two...
       one:  1 wallclock secs ( 0.26 usr +  0.00 sys =  0.26 CPU) @ 96
+1538.46/s (n=250000)
            (warning: too few iterations for a reliable count)
       two:  1 wallclock secs ( 0.46 usr +  0.00 sys =  0.46 CPU) @ 54
+3478.26/s (n=250000)

Benchmark: timing 2000000 iterations of one, two...
       one:  2 wallclock secs ( 2.06 usr +  0.00 sys =  2.06 CPU) @ 97
+0873.79/s (n=2000000)
       two:  4 wallclock secs ( 3.68 usr +  0.00 sys =  3.68 CPU) @ 54
+3478.26/s (n=2000000)
[download]


I don't know too much about the tr/// and s/// code, but it looks like someone went and optimized tr/// between 5.005 and 5.6.1. (Would someone who knows (japhy?) please comment on this?)

I'm pressed for time, but I'd like you to do two things:
1. Try varied lengths of strings, with varied numbers of plus signs.
2. Use a negative count (that is, -5), which tells Perl to run them for that many (positive) seconds.
Please report your findings. I'll be back whenever I can. (I'm talking at TPC today, and I should really be at the talks right now!)

_____________________________________________________
Jeff japhy Pinyan: Perl, regex, and perl hacker.
s++=END;++y(;-P)}y js++=;shajsj<++y(p-q)}?print:??;

Update: Okay, maybe it's not technically faster, but in theory it's supposed to be faster. I'm guessing that the s/// regex command gets more bit-level optimization than anything else in perl, just because it's such a swiss army knife of functionality; maybe tr got left behind on optimization?

"In fact, we must do just the opposite and remain ever-vigilant, striving to ensure that we stop the Urkels of tomorrow before they gain power. Only by demanding full accountability can we reverse the shameful legacy of man's inanity to man."

thanks much :)