Re^2: Supress similar chars in the string

... more rigorously correct ... optimization ...

For me, the issue is not rigor or optimization, but correctness. Insofar as I understand it, the OP requires an overall match on more-than-n contiguous identical characters (although an n-character match may be a component of the overall match). The regexes to which I object match equal-to-or-more-than-n such characters overall, and so are incorrect. The incorrectness of the matches is obscured, but not ameliorated, by the subsequent substitutions, which may replace a string S (which may be empty) with an identical string S.

The point can be exemplified by looking at what is matched by the qr{ (.)\1\1 \K \1* }xms regex (the most elegant, IMHO, in this thread) in a context that does not involve substitution. (Similar examples could be constructed for the other objectionable regexes.) All the empty strings are, in my view, unneeded (and improper) matches.

>perl -wMstrict -le
"my $s = 'aaa...bbbb...ccccc...dddddd.x..yy';
 print qq{'$2'} while $s =~ m{ (.)\1\1 \K (\1*) }xmsg;
"
''
''
'b'
''
'cc'
''
'ddd'
[download]

Anyway, that's my story and I'm sticking to it.

Comment on Re^2: Supress similar chars in the string Select or Download Code

Replies are listed 'Best First'.
Re^3: Supress similar chars in the string by kennethk (Abbot) on Jan 19, 2011 at 20:51 UTC
I would politely disagree with your interpretation, as the two approaches are functionally equivalent. There was no mention of side effects/counting in the OP, and so judging the proposed solution on those grounds is inappropriate. It is an inferior algorithm and likely a bad habit, but the code does the job and does it in a straight-forward fashion. If I'd spent more brain power considering the question, I would have substituted `+` for `*`. The modification actually occurred to me after I posted but before I'd read your comments. However, I try to avoid correcting posts for stylistic reasons alone to avoid versioning confusion and race conditions. I agree that the `\K` construct is the most elegant and appropriate solution, but not everyone runs 5.10 - my external server runs 5.8. The alternate choice option I list seems the most transparent for a neophyte to my eye, but I expect to be the the most expensive given the `e` modifier.	[reply] [d/l] [select]

Replies are listed 'Best First'.

Re^3: Supress similar chars in the string
by kennethk (Abbot) on Jan 19, 2011 at 20:51 UTC

I would politely disagree with your interpretation, as the two approaches are functionally equivalent. There was no mention of side effects/counting in the OP, and so judging the proposed solution on those grounds is inappropriate. It is an inferior algorithm and likely a bad habit, but the code does the job and does it in a straight-forward fashion. If I'd spent more brain power considering the question, I would have substituted + for *. The modification actually occurred to me after I posted but before I'd read your comments. However, I try to avoid correcting posts for stylistic reasons alone to avoid versioning confusion and race conditions.

I agree that the \K construct is the most elegant and appropriate solution, but not everyone runs 5.10 - my external server runs 5.8. The alternate choice option I list seems the most transparent for a neophyte to my eye, but I expect to be the the most expensive given the e modifier.

[reply]
[d/l]
[select]