It starts with lambasting the fact that Ted Dziuba didn't intend the Fibonacci example to be the one role model of comparing Node.js and other languages, and then the Node.js community (or at least a few members of it) rallied around showing that Fibonacci was actually fast in Node. I agree with this. We should be talking about the big picture, not just single implementations.
Then, it goes on for 4 paragraphs about how slow Node.js was with Fibonacci.
Wasn't it the point that Fibonacci wasn't the point? If the point is to show that Node.js is inefficient with computation, and Fibonacci was just a dumb example, why was the rest of the article about the dumb example mentioned previous?
the point was that putting any $computationally_expensive_fn in your event loop was a terrible idea.
in this case he then did
$computationally_expensive_fn = fib;
the point is, unless this event loop pretty much does nothing or hands off the work to a pool and returns immediately, it is utterly unscalable and needs a multiplexer to sit infront of it.
most servers do
while(1) {
$conn = accept_conn();
new EventThread($conn);
}
Thus it's simply multiplexed. Alternately most current other web languages take the approach of integrating with apache which does this one level [up/down?] and they don't have to worry about it at all.
The point is you can write safer code in other languages even if it is slower by it self. With node.js you can do some really dangerous things if you go it alone. You must deploy node with something else, but that's kind of an after thought left up to the dev, and loads won't because they are all under the mistaken impression that because node.js is faster than php/python etc that alone solves all problems and excuses them from sane scalable coding practices. At best it just puts it off a little further down the road. And anything built into a webserver doesn't have to worry, while node will.
I thought the point of this article was to generate a cliff notes version of the last week of arguments, coming from the side of someone who thinks node is cancer. Why are we even discussing this?
Yeah. I really want to see this discussion taken in a direction where we're actually talking about Node.js and its computational (non-?)efficiency, instead of this needless back-and-forth.
It starts with lambasting the fact that Ted Dziuba didn't intend the Fibonacci example to be the one role model of comparing Node.js and other languages, and then the Node.js community (or at least a few members of it) rallied around showing that Fibonacci was actually fast in Node. I agree with this. We should be talking about the big picture, not just single implementations.
Then, it goes on for 4 paragraphs about how slow Node.js was with Fibonacci.
Wasn't it the point that Fibonacci wasn't the point? If the point is to show that Node.js is inefficient with computation, and Fibonacci was just a dumb example, why was the rest of the article about the dumb example mentioned previous?