in reply to Re: Re: Re: should this backspace removal code be done better?
in thread should this backspace removal code be done better?
A matter of style, but I don't care for do 1 while ... since you could just do a while with an empty block. Anyhow, I decided to benchmark all three ways:
#!/usr/bin/perl -w use Benchmark qw(cmpthese); $s = "\bthis is an\b correct\b\b\b usage\b"; sub uk { $a = $s; $a =~ s[(?:[^\cH]\cH|^\cH)][]g while 1+index $a, chr(8); $a; } sub new1 { $a = $s; while ($a =~ s/(?:[^\cH]\cH|^\cH+)//g) {} $a; } sub smack { $a = $s; do 1 while ($a =~ s/(?:[^\cH]\cH|^\cH+)//g); $a; } cmpthese(-5,{uk=>\&uk,new1=>\&new1,smack=>\&smack}); $s x= 100; cmpthese(-5,{uk=>\&uk,new1=>\&new1,smack=>\&smack}); __END__ Benchmark: running new1, smack, uk for at least 5 CPU seconds... new1: 5 wallclock secs ( 5.23 usr + 0.00 sys = 5.23 CPU) @ 31 +024.33/s (n=162393) smack: 6 wallclock secs ( 2.16 usr + 3.05 sys = 5.22 CPU) @ 29 +16.79/s (n=15222) uk: 5 wallclock secs ( 5.19 usr + 0.00 sys = 5.19 CPU) @ 36 +570.60/s (n=189710) Rate smack new1 uk smack 2917/s -- -91% -92% new1 31024/s 964% -- -15% uk 36571/s 1154% 18% -- Benchmark: running new1, smack, uk for at least 5 CPU seconds... new1: 6 wallclock secs ( 5.48 usr + 0.00 sys = 5.48 CPU) @ 36 +8.11/s (n=2016) smack: 6 wallclock secs ( 5.01 usr + 0.41 sys = 5.41 CPU) @ 32 +6.56/s (n=1768) uk: 5 wallclock secs ( 5.23 usr + 0.00 sys = 5.23 CPU) @ 46 +2.14/s (n=2419) Rate smack new1 uk smack 327/s -- -11% -29% new1 368/s 13% -- -20% uk 462/s 42% 26% --
For whatever reason, your version hits sys very hard. However, BrowserUK's version is the fastest of the pack.
Hope this helps.
antirice
The first rule of Perl club is - use Perl
The ith rule of Perl club is - follow rule i - 1 for i > 1
|
|---|
| Replies are listed 'Best First'. | |
|---|---|
|
Re: Re: Re: Re: Re: should this backspace removal code be done better?
by eyepopslikeamosquito (Archbishop) on Oct 05, 2003 at 07:59 UTC | |
by antirice (Priest) on Oct 05, 2003 at 08:27 UTC | |
by shenme (Priest) on Oct 05, 2003 at 10:18 UTC |