From: Jacob Fugal <lukfugl@...>
Date: 2006-10-24T00:04:37+09:00
Subject: Re: DRY fanatics?

On 10/21/06, Ken Bloom <kbloom@gmail.com> wrote:
> On Sun, 22 Oct 2006 10:02:37 +0900, Giles Bowkett wrote:
> >     when /^([A-Za-z0-9,]+), '([^']+)', '([^']+)', '([^']+)'/
>
>  * Use .+? instead of [^']+
>    .+? does a non-greedy match, which is what you're really trying to say
>    with the [^']+

Actually, no, the way he had it is better. Check this out:

http://perlmonks.org/?node=Death%20to%20Dot%20Star!

Using /.+?/ (which is really equivalent to /..*?/ and thus in the same
camp as the article) can be incorrect and also slower. In this case --
I think -- the '.+?' would yield correct results since there's a one
character terminator, but the speed is still an issue. For a simple
string and pair of regexes, it's not much:

  $ cat regex-bm.rb
  require 'benchmark'

  TIMES = 10_000_000
  REGEX1 = /'([^']+)'/
  REGEX2 = /'(.+?)'/
  STRING = "'Woah,' John said, 'there're multiple quotes!'"

  Benchmark.bmbm do |x|
    x.report("[^']+") { TIMES.times{ STRING =~ REGEX1 } }
    x.report(".+?")   { TIMES.times{ STRING =~ REGEX2 } }
  end

  $ ruby -v regex-bm.rb
  ruby 1.8.4 (2005-12-24) [powerpc-darwin7.9.0]
  Rehearsal -----------------------------------------
  [^']+  24.130000   0.050000  24.180000 ( 24.261712)
  .+?    25.490000   0.040000  25.530000 ( 25.799146)
  ------------------------------- total: 49.710000sec

              user     system      total        real
  [^']+  24.160000   0.070000  24.230000 ( 24.729463)
  .+?    25.410000   0.060000  25.470000 ( 25.572987)

But it is present. And it gets worse as both the regex being used and
the string being matched get more complex. For a simple case like
Giles, I wouldn't worry too much about the performance difference
between .+? and [^']+, and the correctness is fine. So you can use .+?
-- it *is* more readable. But it also hides subtleties, which raises
flags for me.

Jacob Fugal