http://toys.lerdorf.com/ Rasmus' Toys Page - en Serendipity 1.7-alpha1 - http://www.s9y.org/ Thu, 01 Jan 1970 00:00:00 GMT http://toys.lerdorf.com/templates/default/img/s9y_banner_small.png http://toys.lerdorf.com/ 100 21 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4698 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (George Cooper) Christian, We've been playing with this concept for a little over a month now. In fact, our lead engineer on this is named Christian too! Pretty funny. If you want to see some of what we're doing, check out https://sourceforge.net/projects/eglooframework on the branch feature/Server. Early benchmarks are impressive. We've been one of the fastest frameworks for a while now and are starting to get traction - Petflow.com uses us for instance - but once we started implementing our stateful context model we saw performance gains of over 1800% w/ Smarty and almost a full 2500% w/ embedded template engine extensions. Check us out. Always interested in feedback. Best, George Fri, 30 Sep 2011 10:35:27 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4698 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4697 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (Rasmus) zmq_poll() doesn't use epoll/kqueue, it is just a regular poll(). zmq does use epoll/kqueue for the internal io threads. However, I am not sure there is much of a performance difference in this case. It might be nice to get some benchmarks on that. Beyond the epoll/kqueue thing, libevent also has a richer featureset and if you already have a libevent-based application it is useful to know that you can slide zmq in there quite easily. Fri, 30 Sep 2011 09:49:19 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4697 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4695 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (Amaury Bouchard) In addition to ZMQ sockets, ZMQ's polling is able to manage any PHP sockets and streams, as well as timeouts. Maybe it's less configurable than libevent, but I never found a real use case where a feature was missing. Libevent is pretty smart (using epoll, kqueue, ...). I don't know how ZMQ's polling is implemented; perhaps there is some systems on which it's less optimized than libevent, but honestly I never saw any problem with it. Am I missing something? Fri, 30 Sep 2011 02:06:08 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4695 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4689 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (Rasmus) If all you are doing is ZeroMQ, perhaps not. But if you are also fielding other types of events then zmq's polling isn't going to help you, and you probably don't want to use both libevent and the built-in polling. In those cases making zmq play nice with libevent becomes quite important. Thu, 29 Sep 2011 14:29:11 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4689 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4688 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (Amaury Bouchard) For this kind of development, is there a real benefit to use libevent instead of ZeroMQ's polling feature? Thu, 29 Sep 2011 14:17:33 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4688 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#c4686 http://toys.lerdorf.com/archives/57-ZeroMQ-+-libevent-in-PHP.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=57 nospam@example.com (Christian Soronellas) Hi Rasmus, Great post! Congrats for it! :D In fact I'm writing an HTTP server based on libevent (for now, is on early development stages) ... but I didn't think to use ZMQ and I just realize that could be interesting to use it ... If want to check it, it's on github! https://github.com/theUniC/Hawk Greetings! Christian. Thu, 29 Sep 2011 00:30:21 -0700 http://toys.lerdorf.com/archives/57-guid.html#c4686 http://toys.lerdorf.com/archives/38-The-no-framework-PHP-MVC-framework.html#c4677 http://toys.lerdorf.com/archives/38-The-no-framework-PHP-MVC-framework.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=38 nospam@example.com (Neil Girardi) This is a great article. If you have a framework that suits you well, use it. Knowing how to "roll your own" will give you a deeper understanding of building applications. The responses are actually more interesting to me than the article itself, for they paint a vivid picture of the obstacles to team work faced by engineering departments. Programmers tend to be somewhat arrogant and fixate on proving that they know more than their peers. Most of the OOP vs. procedural comments and Rails vs. PHP comments have little to do with the article; they are merely attempts by "know-it-all" programmers to prove that they're more clever than Rasmus. If you want to be a great programmer and accomplish awesome things with your dev team you have to have maturity. You have to keep your ego in check and focus on working with your peers, not besting them. My two cents... -n Sat, 20 Aug 2011 13:39:25 -0700 http://toys.lerdorf.com/archives/38-guid.html#c4677 http://toys.lerdorf.com/archives/50-Using-pecloauth-to-post-to-Twitter.html#c4675 http://toys.lerdorf.com/archives/50-Using-pecloauth-to-post-to-Twitter.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=50 nospam@example.com (Dave) Hi, I ran into a small problem with the sending of tweets. I found that when sending tweets the $oauth->fetch command needs to have the 3rd parameter (method) set to "POST" otherwise I got access denied errors :-Dave Fri, 19 Aug 2011 04:35:43 -0700 http://toys.lerdorf.com/archives/50-guid.html#c4675 http://toys.lerdorf.com/archives/56-ASRock-Sandy-Bridge-Motherboard-notes.html#c4638 http://toys.lerdorf.com/archives/56-ASRock-Sandy-Bridge-Motherboard-notes.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=56 nospam@example.com (Andrew S) Though I don't know if this is actually true in Linux, having swap available can increase system performance, as RAM can often be more efficiently used for caching than for infrequently accessed memory. Tue, 24 May 2011 16:38:09 -0700 http://toys.lerdorf.com/archives/56-guid.html#c4638 http://toys.lerdorf.com/archives/56-ASRock-Sandy-Bridge-Motherboard-notes.html#c4637 http://toys.lerdorf.com/archives/56-ASRock-Sandy-Bridge-Motherboard-notes.html#comments http://toys.lerdorf.com/wfwcomment.php?cid=56 nospam@example.com (Rasmus) I have a kill-a-watt. I'll measure them when I get a chance. Mon, 23 May 2011 21:50:17 -0700 http://toys.lerdorf.com/archives/56-guid.html#c4637