|Perl: the Markov chain saw|
Re: Perl is deadby Withigo (Friar)
|on Oct 26, 2006 at 03:22 UTC||Need Help??|
okay, I can't avoid biting this one. :)
While I agree that nearly every point in his blog rant against perl is plain wrong at best, and ridiculous at worst, there might be a solid underlying reason behind the spewing of hatred. This reason seems to be the driving force behind the anti-perl attitude which is far advanced amongst the ranks within Amazon, at least as far as I've seen firsthand.
It's not usually "perl sucks"--everyone sees perl as being as useful and as entrenched as C, so it's not going away anytime soon, and it will always be around in some capacity. And nobody's like "perl coders produce crap", instead it's "perl coders are too smart for everyone else's good, they need to be slowed down so that management can control the situation." The problem raised against perl is always: "we need to setup systems which perform highly specific tasks tens of millions of times per day without fail and perl is just too slow, and the cost of finding perlguts experts to 'do it the right way in perl' is far more expensive than just doing it in Java".
When you have god-knows-how-many systems of servers acting as load balancers doing distributed computation for all kinds of mind-boggling activities and when the same systems in Java require less than half as many servers, then the limits of what can be done in perl have been reached, and a lower level solution is a better fit.
Always use the right tool for the right job, right?
Sure perl is perfect for 90% of whatever you need to do, but when it comes to specialized domains involving highly scalable and highly used systems that don't fail and which involve a touch of scientific simulation and modeling, perl is not the silver bullet. (I hope Perl 6 will be! C'mon Parrot!) Especially once you reach the point of diminishing marginal returns from "throwing more hardware at the problem".
So I can't really blame him for just repeating an attitude he probably heard all over the place at his former job. Though I am a bit puzzled by his recent embracing and promotion of ruby, which falls far short of perl in terms of scalability and performance.
Alas, such is the IT industry, always creating new bridges to sell you; which are worse in terms of usability, worse in using the newest hardware more inefficiently and worse in providing inumerable new and useless features that only increase the overall complexity in order to justify the vicious circle of profit.
But nobody ever got fired for buying the new-new-thing (and trash talking the old thing). In fact, that's how you get promoted!