Re: User Existance?

Well, let's benchmark and see. I created a /etc/passwd file with over 6000 users each named x and some number. Here I search for the user x1, x50, x1000, x3000, and x6000. I test with getpwnam and a regex. Each test is tried 1000 times and the output follows(All tests run on a PII-450 with 384MB RAM running Redhat 9):

#!/usr/bin/perl -w

use Benchmark;

for $x (1, 50, 1000, 3000, 6000){

timethese (1000,{
        "getpwnam${x}" => q {
        $there=getpwnam("x${x}")
        },
        "regex${x}" => q {open(PASSWD,"/etc/passwd");
        while(<PASSWD>)
        { if (/^x${x}:/){$alsothere=1};
          last if $alsothere == 1;
        }
        close PASSWD;},
});
}



Benchmark: timing 1000 iterations of getpwnam1, regex1...
 getpwnam1:  0 wallclock secs ( 0.13 usr +  0.05 sys =  0.18 CPU) @ 55
+55.56/s (n=1000)
            (warning: too few iterations for a reliable count)
    regex1:  0 wallclock secs ( 0.10 usr +  0.03 sys =  0.13 CPU) @ 76
+92.31/s (n=1000)
            (warning: too few iterations for a reliable count)
Benchmark: timing 1000 iterations of getpwnam50, regex50...
getpwnam50:  1 wallclock secs ( 0.29 usr +  0.02 sys =  0.31 CPU) @ 32
+25.81/s (n=1000)
            (warning: too few iterations for a reliable count)
   regex50:  0 wallclock secs ( 0.08 usr +  0.05 sys =  0.13 CPU) @ 76
+92.31/s (n=1000)
            (warning: too few iterations for a reliable count)
Benchmark: timing 1000 iterations of getpwnam1000, regex1000...
getpwnam1000:  2 wallclock secs ( 2.36 usr +  0.24 sys =  2.60 CPU) @ 
+384.62/s (n=1000)
 regex1000:  0 wallclock secs ( 0.10 usr +  0.04 sys =  0.14 CPU) @ 71
+42.86/s (n=1000)
            (warning: too few iterations for a reliable count)
Benchmark: timing 1000 iterations of getpwnam3000, regex3000...
getpwnam3000:  8 wallclock secs ( 7.26 usr +  0.43 sys =  7.69 CPU) @ 
+130.04/s (n=1000)
 regex3000:  1 wallclock secs ( 0.12 usr +  0.05 sys =  0.17 CPU) @ 58
+82.35/s (n=1000)
            (warning: too few iterations for a reliable count)
Benchmark: timing 1000 iterations of getpwnam6000, regex6000...
getpwnam6000: 16 wallclock secs (14.73 usr +  0.67 sys = 15.40 CPU) @ 
+64.94/s (n=1000)
 regex6000:  0 wallclock secs ( 0.13 usr +  0.06 sys =  0.19 CPU) @ 52
+63.16/s (n=1000)
            (warning: too few iterations for a reliable count)
[download]

Suprisingly the regex wins hands down. Not too shabby. I don't think you need to worry about your script being slow at all.

HTH

Comment on Re: User Existance? Download Code

Replies are listed 'Best First'.
Re: Re: User Existance? by pzbagel (Chaplain) on May 17, 2003 at 18:50 UTC
I apologize, my while loop exit code was wrong. No wonder the regex beat getpwnam. Here is the correct code and results. Stick wirh getpwnam(). #!/usr/bin/perl -w use Benchmark; for $x (1, 50, 1000, 3000, 6000){ timethese (1000,{ "getpwnam${x}" => q { $there=getpwnam("x${x}") }, "regex${x}" => q {open(PASSWD,"/etc/passwd"); while(<PASSWD>) { if (/^x${x}:/){$alsothere=1; last}; } close PASSWD;}, }); } Benchmark: timing 1000 iterations of getpwnam1, regex1... getpwnam1: 0 wallclock secs ( 0.12 usr + 0.06 sys = 0.18 CPU) @ 55 +55.56/s (n =1000) (warning: too few iterations for a reliable count) regex1: 0 wallclock secs ( 0.41 usr + 0.02 sys = 0.43 CPU) @ 23 +25.58/s (n =1000) Benchmark: timing 1000 iterations of getpwnam50, regex50... getpwnam50: 1 wallclock secs ( 0.28 usr + 0.02 sys = 0.30 CPU) @ 33 +33.33/s (n =1000) (warning: too few iterations for a reliable count) regex50: 1 wallclock secs ( 0.80 usr + 0.04 sys = 0.84 CPU) @ 11 +90.48/s (n =1000) Benchmark: timing 1000 iterations of getpwnam1000, regex1000... getpwnam1000: 2 wallclock secs ( 2.41 usr + 0.20 sys = 2.61 CPU) @ +383.14/s ( n=1000) regex1000: 9 wallclock secs ( 8.52 usr + 0.13 sys = 8.65 CPU) @ 11 +5.61/s (n= 1000) Benchmark: timing 1000 iterations of getpwnam3000, regex3000... getpwnam3000: 9 wallclock secs ( 7.36 usr + 0.38 sys = 7.74 CPU) @ +129.20/s ( n=1000) regex3000: 28 wallclock secs (24.94 usr + 0.54 sys = 25.48 CPU) @ 39 +.25/s (n=1 000) Benchmark: timing 1000 iterations of getpwnam6000, regex6000... getpwnam6000: 16 wallclock secs (14.44 usr + 0.77 sys = 15.21 CPU) @ +65.75/s (n =1000) regex6000: 50 wallclock secs (49.24 usr + 1.17 sys = 50.41 CPU) @ 19 +.84/s (n=1 000) [download] Forgiveness Please. Beating myself with a bamboo cane as I speak.	[reply] [d/l]
Re: Re: User Existance? by Anonymous Monk on May 17, 2003 at 18:23 UTC
Thanks for that. Using the regexp method, how would i apply it to this custom made file: `user1 user2 user3 user4 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5` [download] As you can see, the file varies slightly. So i want the regexp to open up the file (located at /etc/adduser/user.log) and match $username against the first part (userX) of each line. If there is a match, print an error. Cheers peeps, The help is massivly appreciated :)	[reply] [d/l]
Re: Re: Re: User Existance? by The Mad Hatter (Priest) on May 17, 2003 at 18:52 UTC
Probably a better way to do this, but here's one way: Update So `getpwnam` is faster? Ok... `my $username = "usernamehere"; print "Duplicate!\n" if getpwnam($username);` [download] `my $username = "user1"; while (<DATA>) { chomp; my @info = split ","; foreach (@info) { print "Duplicate!\n" if /\A\Q$username\E\z/ } } __DATA__ user1 user2 user3 user4 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5 user5,pass5,email5,ip5` [download]	[reply] [d/l] [select]
Re: Re: Re: Re: User Existance? by Anonymous Monk on May 17, 2003 at 20:37 UTC
why am i getting an error when trying to run this? `if getpwnam($username) { $message = $message.'<p>The Chosen Username Already Exists.</p>'; $found_err = 1; }` [download]	[reply] [d/l]
Re: Re: Re: Re: Re: User Existance? by The Mad Hatter (Priest) on May 17, 2003 at 20:40 UTC