Is there a particular reason that Zend_Uri_Http uses PCRE for its underlying implementation instead of basically just wrapping parse_url()? As far as I can tell, the current implementation and parse_url() are equally capable of returning the data fragments that the class makes available (and correct me if I'm wrong on this). The current implementation has two main disadvantages over parse_url(): 1) the current implementation requires the PCRE extension which, though enabled by default, may not be available to (or needed by) all users, whereas parse_url() is part of the core; 2) the current implementation is in PHP whereas parse_url() is in C, so I'm willing to bet there's a performance advantage (albeit one that's not extremely substantial, but I don't have any figures to back that up). I would imagine there is a good reason that I'm just not seeing and I'm just curious to see what it is. If there's no good reason not to, I'd be willing to come up with an alternate implementation and run some benchmarks and report my findings on the list.
Regards, Matthew Turland