On Wed, Aug 19, 2009 at 12:28 PM, Rickard Öberg<[email protected]> wrote: > Niclas Hedhman wrote: >> >> IDEA can easily be taught (alt-enter on warning) that an annotation is >> used for dependency injection. I don't know Eclipse at all... >> >> I think I am suggesting that we keep the way it is and can re-evaluate >> this feature later. > > Well... I have now spent about two days on this, and have everything > working, with the same performance (or better) as before. > > The main "gotcha" is that Qi4j is now pickier with that you can't use > "final" on fragment classes, as they need to be subclassed, and that they > have to be marked as public (same reason). But the feature itself I think is > very handy to have, and makes the code easier to read and write, so I would > like to commit it anyway. > > WDYT?
Since this approach gives exactly 1 extra subclass per Fragment type, independently of how many times it is used in Composites, I think we have a 'reasonable' overhead. It will however change the claim of "No Magic under the Hood", and definitely stop corner use-cases such as Android and potentially even Google AppEngine and other future cloudservices, which most likely will block bytecode manipulation. So, I am +1 to making this enhancement now, and that we in the longer term try to look for a solution where we can break this functionality out to compile time so that we can manage cloud services, Android and other restricted environments (Applets?). Cheers -- Niclas Hedhman, Software Developer http://www.qi4j.org - New Energy for Java I live here; http://tinyurl.com/2qq9er I work here; http://tinyurl.com/2ymelc I relax here; http://tinyurl.com/2cgsug _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
