On Mon, Jul 23, 2012 at 10:32 AM, Per <per.fragem...@gmail.com> wrote: > > I doubt that Google will commit to any official statement. So here's my 2 > cents, from a low-usage, rather complex B2B application. "Frameworks are > okay, to a certain degree, but be prepared to rework if you're successful"
While I certainly would not disagree with this, I think the important discussion - the one relevant to the question of whether Google should spend effort refining the scheduler - is "what kind of frameworks are reasonable?" Let's take persistence. Does anyone honestly believe that it's reasonable to write complicated applications using only the Java low-level api? So you pick a lightweight persistence framework like Objectify. Great choice! ;-) I'd like to think I've done a pretty good job of optimizing Ofy, but there is a problem I do not have any way of getting around: In order to map POJOs to the datastore, Objectify has to introspect the POJOs. Because queries can return polymorphic entities, this introspection has to happen up front. Everything is great when your app has a handful of classes. But you add features and add features and now you have 30, 50, 100+ entity classes. My domain model currently has 56 classes and it takes Objectify 4-5s to load and introspect them. This isn't a classpath scanning issue - the bottleneck seems to be the classloader itself. My point: The more complicated your application is, the more startup time is going to be impacted by a framework like Objectify. Also, the more complicated your application is, the more you need a framework like Objectify. It's a catch-22. So this 4-5s hit is just something I have to accept. There's no magic "optimization" I can do to make it go away - believe me, I'm in the best position of anyone to do it, and I don't see any fixes for Objectify that preserve important features like polymorphism. This hit is going to get worse when my domain model doubles, triples, etc over the next few years. If we say "all apps must start up in <5s" then we've basically said that Brandon is right and we must stop using all frameworks, even at the foundational level of persistence. I certainly hope nobody takes that seriously. I'm not saying that it's a great idea to throw in frameworks that make your app take 50s to start up... but I am saying that Java startup latency is not a problem we can optimize away. It's something we have to deal with one way or another, and it would be best for everyone if Google dealt with it at the scheduler. Jeff -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to google-appengine@googlegroups.com. To unsubscribe from this group, send email to google-appengine+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
