Re: Testing a string for a range of characters

Just another WTDI:

# predefine $testlist as a simple string of the set of acceptable char
+acters then
  'index' => sub {
                   $test_char = chr( int( rand 256 ) );
                   if (index($testlist, $test_char) > -1 && length($te
+st_char)==1) {} else {}
                 },
[download]

which gives the surprisingly (to me) competitive result:

Benchmark: timing 10000000 iterations of comp, exist, index, index2, r
+egex, regex2, regex3...
      comp: 44 wallclock secs (41.64 usr +  0.00 sys = 41.64 CPU) @ 24
+0147.93/s (n=10000000)
     exist: 41 wallclock secs (38.30 usr +  0.00 sys = 38.30 CPU) @ 26
+1117.06/s (n=10000000)
     index: 42 wallclock secs (41.30 usr +  0.00 sys = 41.30 CPU) @ 24
+2148.34/s (n=10000000)
    index2: 40 wallclock secs (39.48 usr +  0.00 sys = 39.48 CPU) @ 25
+3260.73/s (n=10000000)
     regex: 55 wallclock secs (53.28 usr +  0.00 sys = 53.28 CPU) @ 18
+7684.17/s (n=10000000)
    regex2: 53 wallclock secs (52.34 usr +  0.01 sys = 52.36 CPU) @ 19
+0989.13/s (n=10000000)
    regex3: 53 wallclock secs (50.97 usr +  0.00 sys = 50.97 CPU) @ 19
+6197.69/s (n=10000000)
[download]

I also tested a couple of variations:
regex2 - used [A-Z] in place of [A..Z] as I've never used the latter before and can't even figure what it would do at the moment
regex3 - omitted the ^ and $, which violated the spec but I wanted to know
index2 - similarly omitted the length() call

Update: Benchmarks run on NT4 PIII-500 w/256MB RAM, Perl 5.6.0, ActiveState build 618

--
I'd like to be able to assign to an luser

Comment on Re: Testing a string for a range of characters Select or Download Code

Replies are listed 'Best First'.

Re: Re: Testing a string for a range of characters
by BoredByPolitics (Scribe) on Jan 24, 2001 at 21:44 UTC

regex2 - used A-Z in place of A..Z as I've never used the latter before and can't even figure what it would do at the moment

Not what I'd thought it would do! OK, I made a wrong assumption (getting mixed up between array ranges and regex ranges) - you're quite correct, /$[A-Z]^/ is functionally correct.

I tried your suggestion along side the (now correct) regex and exist, with the following results -

Benchmark: timing 10000000 iterations of exist, index, regex...
     exist: 50 wallclock secs (49.39 usr +  0.01 sys = 49.40 CPU)
     index: 61 wallclock secs (60.64 usr +  0.00 sys = 60.64 CPU)
     regex: 85 wallclock secs (83.18 usr +  0.03 sys = 83.21 CPU)
[download]

It's interesting that I get a bigger difference between the exist and index versions. I wonder what OS & hardware you're running on? Mine is linux v2.2.18 on a Pentium II 400Mhz with 128Mb RAM.

Pete

[reply]
[d/l]
[select]


XP is just a number
	PerlMonks