I did some consulting a few months ago. A FOAF had developed and launched a quite comprehensive online textbook, and their success was choking the the website. They don't know much programming, and had they not used a framework (and one of the slower ones at that), they wouldn't have launched. Period.<p>I moved in, got them off shared hosting, replaced the built-in search with a Apache Solr, tuned the cache and a few other quick wins, took my money and left them my phonenumber. They live to fight another year or two, and I'll stop by and help again then.<p>Scalability is not your problem, getting people to give a shit is.<p><a href="http://teddziuba.com/2008/04/im-going-to-scale-my-foot-up-y.html" rel="nofollow">http://teddziuba.com/2008/04/im-going-to-scale-my-foot-up-y....</a><p>Frameworks helps getting people to give a shit, custom C++ extensions doesn't, so use a framework, unless Yahoo! called.
I can't find any further details on this claim. It wouldn't surprise me if it was true but benchmarks really do depend on the details. I want more info.<p>Skipping over the use of "Hello World!" in benchmarks is anyone actually surprised by this? Frameworks aren't a silver bullet. I don't think they were ever meant to be, especially when it comes to performance. There is nothing stopping you from going in to the framework code and optimising it or even taking functionality over to C++ if that's what you want.
So we should all abandon our frameworks entirely for C++ because we're going to be getting Yahoo levels of traffic, trees are people too, and no one's ever achieved 280 req/sec with a framework because things like caching don't exist.<p>And the semantic web is cool.<p>What the fuck? Why is anyone upvoting this link? It's completely insane.
Ah, Rasmus. Rasmus is kind of like the last hockey player who doesn't wear a helmet and looks down on the rest of us pussies. We don't all have 500 million users, man.<p>He's a got a good point: PHP <i>is</i> a framework. It's just not a very easy one. Clue, Ras: when your users ask you constantly for certain features (say, named parameters or templating) and you don't give it to them, they will go around you. Or they'll stop using your language.
What a load of crap. Quite surprising to hear that sort of tripe come from a respected figure like Rasmus Lerdorf... but then, he did invent PHP.<p>First of all, requests/s per server is not a measure of performance, it's a measure of scalability. I can take the same system he's got, buy 9 other identical servers, and get 10 times the throughput. Does that mean they're higher performance? No, just higher scalability.<p>Secondly, scalability is not your problem. Getting people to give a shit about your site/product is your problem. Scalability is a hard problem later, but it is only worth really worrying about once it is a real problem.<p>Rasmus Lerdorf: -1.
"To get the speed that is necessary for truly massive web systems you have to use compiled C++ extensions to get true, scaleable architecture."<p>But at the same time, even creating a simple procedural PHP extension is a huge undertaking (much more than it needs to be), accentuated by poor and out of date documentation all around, and large parts of documentation are missing from the PHP site concerning extension creation.