Monday December 08, 2008

The other day, I was patching some Perl code. There I was, in the zone, code streaming off of my finger tips. But wait, I was writing Python in the middle of a Perl subroutine. Um. I found the bare word and missing semi-colon errors invoking perl -cw dingleberry.pl amusing. How did that happen?

Truth is, I find myself rarely using Perl anymore. I spent years building applications with Perl. Having made extensive use of mod_perl APIs and various CPAN modules, studied the Talmudic wisdom of Damian Conway, struggled with the double-edged sword of TMTOWTDI and rolled my eyes at the Perl haters for their failure to appreciate the strange poetry that is Perl it'd seem like a safe bet that Perl would remain on my top shelf. A lot of the complaints of Perl haters seem superficial ("ewe, all of those punctuation characters", shaddup). Yet, Perl has been long in the tooth, for a long time. I recall 5 years ago thinking that Perl 6 wasn't too far away (after all, O'Reilly published Perl 6 Essentials in June 2003). I'm sorry, my dear Perl friends, insistence that Perl is Alive rings hollow, now.

I still find it heartening to hear of people doing cool things with Perl. David McLaughlin's uplifting How I learnt to love Perl, waxing on about Moose and other "modern" Perl frameworks (but come on dude, everyone knows that PHP, sucks, heh). Brad Fitzpatrick released Perl for Google App Engine today. But Perl, I'm sorry. It's just too little, too late.

I'm just weary of the difficulties achieving team adherence to disciplined coding practices (or even appreciate why they're especially necessary in the TMTOWTDI world of Perl). The reliance on Wizardry is high with Perl; the path from novice to master requires grasping a wide range of arcana. Is it too much to ask for less magic in favor of easier developer ramp up? Perl's flexibility and expressiveness, it's high virtues, also comprises the generous reel of rope that programmers routinely hang themselves with. On top of idiomatic obscurities are the traps people fall into with dynamic typing and errors that only make themselves evident at runtime. Good testing practices are usually the anecdote to the woes of dynamic typing and and yet writing a good test harness for a Perl project is often a lot of work compared to the amount of work required to write the application code.

I well understand the security that programmers feel using static typing but I'm not saying the static typing is the cure to any ills. The compiler is the most basic test that your code can be understood and it gives your IDE a lotta help. That's great but static typing is also an anchor dragging on your time. From what I can tell, Java is the new C/C++ and Jython, JRuby. Groovy and friends (Scala, Clojure, etc) are the ways that people program the JVM with higher productivity. I'm not saying fornever to Perl (that's a long time). But I am saying Hasta La Vista, for now. I've been quite productive lately with Python (I know Perl friends, heresy!) and plan on pushing ahead with that, as well as with Java and other JVM languages. And where necessary, using Thrift to enable the pieces to work together. Python is certainly not perfect, it's quirks are many, too. But I've seen recent success with collaborative software development with Python that would have been difficult to accomplish with Perl. I'm not trying to stoke any language war at all, I'm just reflecting on how I've drifted from Perl. Amongst people I know and things I've read elsewhere, I'm not the only one. Don't fret, Perl, I'm sure I'll see you around.

perl   python   java   programming