Blogs : Latest entries
|
|
|
Last night I upgraded all important software on the server of one of our bigger customers. We write all their applications with RIFE and the following sites run on that one machine (all with dedicated admin sections and all public pages being fully dynamic): Netwerk, Dag Allemaal, Genieten, Goed Gevoel, Magnet Magazines, Netcetera, TV-2Weken, TV Familie and Uw Vermogen. The new software and versions are: The performance is now really outstanding. Connections are responding instantly, even when a lot of images need to be loaded at once (for instance on the netwerk.be homepage). Note that all dynamic images (magazine covers, avatars, news posts, ...) are served straight out of the Postgres database as a stream. We don't write any lose files to disc from the content management interface. I'm really happy about this software combination and will start looking into the JRockit management console in a little while. Maybe it can provide some interesting information. |
|
I had some conversions with Obie during the last days about many web-related things (like Laszlo, RoR, Java web frameworks, ...). I didn't know him before, but he seems to be an interesting person and one of the few who you can really discuss something constructively on IRC with. In his latest blog entry he manages to explain one of the problems in Java-framework-land very well: there's so much out there that people stopped taking the trouble to look at alternatives that don't have a high profile, and he points to RIFE as one project that people should take a look at. Thanks, I agree! There's nothing wrong with this approach, I stopped looking at every new thing too. However, in communities like Ruby, competition is much less. This makes it easier for people to follow new releases and actually try out what others come up with. I think that the Java community is nowadays getting divided in two major camps: those that base themselves on solid, proven methods (mostly enterprises) and those that are looking at ways to improve their day-to-day work by adopting or coming up with creative solutions to problems. However the people I know from the second category already went through the long process where you learn the hard way why basic things like templates without scriptlets and separation of flow and implementation matter. I for one, am prepared to have an overhead of a couple of hundred lines of centralized setup files to get a clearly structured system that's very maintainable over time. I personally think this is even important is very small applications, since you start on a solid foundation that doesn't prevent you from growing (and most of the small things we made, grew). However, in one of his previous blog entries, Obie states that all Java web frameworks suck. This is something I totally don't agree with since I personally think RIFE is great! When I look through the alternatives I even see others with great potential but with another focus (like Wicket). A pattern that I've however witnessed in many occasions is that people take a glance at the XML version of RIFE's site structure and immediately decide that it's just another action mapping framework and move on. This is probably totally our own fault, since we clearly seem to lack the marketing skills of the RoR crowd. We will thus now try to ramp up our PR skills a bit and look at ways to convince people to at least take a look at our stuff with some knowledge of the advantages ... stay tuned! |
