Front page | perl.perl5.porters |
Postings from February 2009
2009/2/24 Aristotle Pagaltzis <pagaltzis@gmx.de>: > * demerphq <demerphq@gmail.com> [2009-02-24 13:15]: >> All this ranting over File::Find is IMO highly unconvincing. >> >> It works, it does whats needed out of the box for most >> purposes, and frankly the replacments ive seen tend to be >> harder to use, harder to understand and just another silly >> dependency that i remove when i encounter (because most times >> i encounter it i dont have the silly dependency handy and >> replacing it faster than finding the dependency, especially >> when you are on a train with no connectivity. > > As far as I’m concerned this is not about replacing File::Find > with a more declarative API. > > My real beef with FF is that it’s callback-based. Well i can see how an iterated FF would be useful, but not so useful as to deprecate File::Find. Although possibly useful enough to rewrite FF so it is internally iterative and then use the iterative version to implement the callback framework. But i wonder about how much state you might have to store for an iterative traversal. If you were unlucky this might require something close to reading all the files anyway. Im wondering if iterative and callback based designs have different system costs associated to them. For instance i can imagine at least attempting an argument like "callback designs are less likely to be vulnerable to memory issues resulting from ridiculously large directories". (Not saying id win the argument mind you.) > Another rather minor beef is that it’s annoyingly verbose to just > get a list of files when that is what I want – minor because this > can be fixed with a wrapper Seems like this should be a utility routine in FF and not an excuse to eliminate FF. >(callback-vs.-iterator cannot). No. True. But i still would prefer to see effort go into modifying FF and extending it than to replacing it. > > Hmm, I wonder how much of a problem it would be to fix those > instead of replacing FF outright… FF is some really ugly code > anyway… That is true. But it also works in a lot of environments with different file systems and rules. Thus it is not surprising that it is crufty, and arguable that anything with similar functionality is going to be just as crufty. Cheers, yves -- perl -Mre=debug -e "/just|another|perl|hacker/"Thread Previous | Thread Next