All the Perl that's Practical to Extract and Report

Regular expressions

gnat on 2001-12-16T20:29:28

I've been documenting POSIX-style regular expressions for The PHP Book, and they're starting to grow on me. Perl has some nice stuff ($ matching before trailing newline) but the sheer readability (and extensibility) of the POSIX named character classes (e.g., [[:cntrl:]]) is very nice. I know Perl now supports POSIX classes, but I get the feeling they're thought-of as second-class citizens.



My current "oh my GOD" reaction to PHP is the great inconsistency in the function names. String functions start with str_, string_, and sometimes something else. This is important, of course, because PHP doesn't have namespaces. The POSIX regular expression function names are only somewhat similar to the Perl regular expression function names. And so on.



This really shows up the dangers of unplanned growth. I'm not suggesting design by committee, but simply articulating a few module design principles could have solved these messes.



Which principles? I'm thinking things like:

  1. Simple is better
  2. Look at other similar modules
  3. Be consistent
  4. Name similar functions similarly


I suppose 3 and 4 are really the same. That's the problem with trying to write down commonsense. The underlying principles that govern our behavior are sometimes completely invisible to us. That's why we need people like Damian, who can describe what we're supposed to do in such a way that we go "oh! well, duh. Why didn't I think of that myself?"



--Nat

PHP function names

vsergu on 2001-12-17T04:06:04

My current "oh my GOD" reaction to PHP is the great inconsistency in the function names.

You're not kidding. It's great that we can remember that Perl builtins never have underscores. With PHP, the use of underscores is totally random (sort of like CGI.pm used to be). is_array(), is_numeric(), etc., have underscores, but isset() doesn't. strlen(), strtr(), and a bunch of other str* function have no underscore, but str_pad(), str_repeat(), and str_replace() do. And you've got stuff like frenchtojd() taking up the namespace -- for handling the French revolutionary calendar (which of course comes up all the time in my Web work).