Re: Match upper and lower case

IMHO TMTOWTDI

if (uc $error eq "ERROR"){
   # . . .
}
[download]

if (lc $error eq "error"){
   # . . .
}
[download]

Although, I would prefer Zaxo's first example because it seems more efficient (if for no other reason, but conservation of code :-)

-Jim

Comment on Re: Match upper and lower case Select or Download Code

Replies are listed 'Best First'.
Re^2: Match upper and lower case by blazar (Canon) on Jan 30, 2005 at 15:04 UTC
IMHO TMTOWTDI `if (uc $error eq "ERROR"){ # . . . }` [download] or `if (lc $error eq "error"){ # . . . }` [download] Although, I would prefer Zaxo's first example because it seems more efficient (if for no other reason, but conservation of code :-) Note that this is not strictly equivalent to Zaxo's code since `/error/i` checks (case insensitively) for the presence of the string 'error' and not for equality to it (up to case insentiveness). More importantly: "more efficient"? Well, I premise that I'd probably use `/error/i` or `/^error$/i` (this really depends on what can there be in `$_`) myself in most reasonable cases, but running `#!/usr/bin/perl use strict; use warnings; use Benchmark qw/:all :hireswallclock/; my $test_str='error'; my @test=map { my @chr=("\0", " "); local $_=$test_str; s/./$& ^ $chr[rand 2]/ge} 1..100; no warnings 'void'; cmpthese -30, { regex => sub { /^\Q$test_str\E$/i for @test }, lc => sub { $test_str eq lc for @test} }, 'all'; __END__` [download] I get `Benchmark: running lc, regex for at least 30 CPU seconds... lc: 34.068 wallclock secs (34.10 usr 0.00 sys + 0.00 cusr 0 +.00 csys = 34.10 CPU) @ 19917.24/s (n=679178) regex: 31.518 wallclock secs (31.53 usr 0.00 sys + 0.00 cusr 0 +.00 csys = 31.53 CPU) @ 13689.50/s (n=431630) Rate regex lc regex 13690/s -- -31% lc 19917/s 45% --` [download] and `Benchmark: running lc, regex for at least 30 CPU seconds... lc: 31.4462 wallclock secs (31.45 usr 0.00 sys + 0.00 cusr +0.00 csys = 31.45 CPU) @ 26138.57/s (n=822058) regex: 31.8579 wallclock secs (31.86 usr 0.00 sys + 0.00 cusr +0.00 csys = 31.86 CPU) @ 14326.49/s (n=456442) Rate regex lc regex 14326/s -- -45% lc 26139/s 82% --` [download] under Windows 98 and Linux (kernel 2.6.10) respectively. Perl 5.8.6 in both cases...	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^2: Match upper and lower case
by blazar (Canon) on Jan 30, 2005 at 15:04 UTC

IMHO TMTOWTDI
if (uc $error eq "ERROR"){ # . . . }
[download]
or
if (lc $error eq "error"){ # . . . }
[download]
Although, I would prefer Zaxo's first example because it seems more efficient (if for no other reason, but conservation of code :-)

strictly

/error/i

More importantly: "more efficient"? Well, I premise that I'd probably use /error/i or /^error$/i (this really depends on what can there be in $_) myself in most reasonable cases, but running

#!/usr/bin/perl

use strict;
use warnings;
use Benchmark qw/:all :hireswallclock/;

my $test_str='error';
my @test=map {
    my @chr=("\0", " ");
    local $_=$test_str;
    s/./$& ^ $chr[rand 2]/ge} 1..100;

no warnings 'void';
cmpthese -30, { regex => sub { /^\Q$test_str\E$/i for @test },
                lc => sub { $test_str eq lc for @test} }, 'all';

__END__
[download]

Benchmark: running lc, regex for at least 30 CPU seconds...
        lc: 34.068 wallclock secs (34.10 usr  0.00 sys +  0.00 cusr  0
+.00 csys = 34.10 CPU) @ 19917.24/s (n=679178)
     regex: 31.518 wallclock secs (31.53 usr  0.00 sys +  0.00 cusr  0
+.00 csys = 31.53 CPU) @ 13689.50/s (n=431630)
         Rate regex    lc
regex 13690/s    --  -31%
lc    19917/s   45%    --
[download]

Benchmark: running lc, regex for at least 30 CPU seconds...
        lc: 31.4462 wallclock secs (31.45 usr  0.00 sys +  0.00 cusr  
+0.00 csys = 31.45 CPU) @ 26138.57/s (n=822058)
     regex: 31.8579 wallclock secs (31.86 usr  0.00 sys +  0.00 cusr  
+0.00 csys = 31.86 CPU) @ 14326.49/s (n=456442)
         Rate regex    lc
regex 14326/s    --  -45%
lc    26139/s   82%    --
[download]

[reply]
[d/l]
[select]