^([-_a-z]+)\.([^/]+)$ %1/$1/$1.$2

[mdlbear]

Yes, as a matter of fact, Apache rewrite rules do look like a combination of line noise and swearing. This is a problem why?

Comments

[selenesue.livejournal.com]

Oh, here I thought you were cussing like in a comic strip.

[mdlbear]

That was what I noticed when trying to point out some of the fine points to Kat, who has no interest in such things.

Most of my comic-strip profanity came yesterday while I was writing the darned thing. Tricky.

[victorthecook.livejournal.com]

To answer your rhetorical question literally -- it's a problem because finding the blasted bugs in my regular expression is made harder thereby.

On the other hand, the widespread adoption of Perl-style regex syntax at least brings some consistency to the problem.

[mdlbear]

Not entirely clear whether verbosity beats compactness for readability. Spread a one-liner out over a full page of function calls and it's not clear whether it's improved thereby.

[victorthecook.livejournal.com]

Yep, it's a trade-off. In Perl per se, you can add whitespace and comments to a regex with an alternate syntax -- but you rarely see it, because it means using an alternate syntax.

That said, the existence of (many) interactive regex builders implies that something about regexes is difficult to do correctly. The question is: is it the thing itself which is difficult, or is the regex language providing (superficial, fixable) friction?

[aerowolf.livejournal.com]

If you look at the problem domain, you're looking at an attempt to describe a language (defined as 'a sequence of characters of arbitrary size') with another.

The design of regular expressions (or at least the process they went through) is really quite elegant. However, elegance sucks when it's impossible to understand.

Of course, one could look at the source of the necessity to describe one language with another as being broken... but it's The Way Things Are. regexps are not the only attempt to define such a language, by the way.