Now that PHP 6 is in the works, there is even less excuse than existed previously for Web sites (hosting providers in particular) not migrating to PHP 5 from PHP 4. We are faced with the unpleasant possibility for tool and library developers of having to support three major, necessarily incompatible, versions of PHP.
I am not yet up to speed on what PHP 6 is going to bring to the table, but PHP 5 (which will be two years old on 13 July 2006) makes PHP a much more pleasant, usable language for projects large and small. With a true object model, access control, exception handling, improved database support, improved XML support, proper security design concepts, and so on, it's a far cry from the revised-nearly-to-the-point-of-absurdity PHP 4.
Another great thing about PHP 5, if not strictly part of it, is the PHPUnit unit testing framework (see also the distribution blog). This is a wonderful tool for unit testing, refactoring, and continuous automated verification of your codebase. It will strongly encourage you to make your development process more agile, using a test first/test everything/test always mindset that, once you have crossed the chasm, will benefit a small one- or two-man shop at least as much as the large, battalion-strength corporate development teams that have to date been its most enthusiastic audience.
I have so far used this tool and technique for three customer projects: the first was delivered (admittedly barely) on time, the second was actually deliverable less than 2/3 of the scheduled calendar time into the project (allowing for further refactoring to improve performance) and delivered on time, and the third was delivered 10% ahead of time, with no heroic kill-the-last-bug all-night sessions required.
Discussing the technique with other developers regarding its use in PHP and other languages (such as Python, Ruby, C++ and of course Java; the seminal "JUnit" testing framework was written for Java), gives the impression that this experience is by no means unique or extreme (nor did I expect it to be). Given that two of my three major career interests for the last couple of decades have been rapid development of high-quality code and the advancement of practices and techniques to help our software-development craft evolve towards a true engineering discipline, this would seem a natural thing for me to get excited and evangelical about. (The third, in case you're wondering, is the pervasive use of open standards and non-proprietary technologies to help focus efforts on true innovation).
All of this may seem a truly geeky thing to rave about, and to a certain degree, I plead guilty of that. But it should also be important, or at least noteworthy, to anybody whose business or casual interests involve the use of software or software-controlled artifacts like elevators and TiVo. By understanding a little bit about how process and quality interact, clients, customers and the general-user public can help prod the industry towards continuous improvement.
Because, after all, "blue screens" don't "just happen".