on 11/27/01 9:39 PM, "Fedor Karpelevitch" <[EMAIL PROTECTED]> wrote:
> I agree that templates would have to implement what is currently done by > Village, but the code would not be equivalent. The reason is that village > does everything in a generic manner, for the reason that it has no knowledge > of the data structures it is working with. But peers do know the structure of > the data they are working with, the types and so on at code generation time, > so that they can resolve most things at that time > > Huge overhead Please create a test case which can be run on a new JVM (ie: 1.3) that shows that the overhead incurred by Village is enough to warrant not using it and to increase the code complexity of Torque and risk introducing bugs. Village has been around since March 1999 and no one has been able to prove that yet in a reliable test case. I think that there are other lower hanging fruit optimizations which will have more impact than removing Village that we can do...such as adding an intelligent caching system to Torque objects. thanks, -jon -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
