Re: JDBCPersistence Technology Preview Release

Chris Uppal wrote:
> alexr wrote:
>
> > Code generation occurs at runtime, so the setup is very easy, there is
> > no build step. SQL is leveraged as the query language.
>
> Seems a sensible approach. I'm surprised that it's not more widely used.
>
> After all, Sun's class library uses bytecode generation in several places (to
> implement reflective method invocation, for instance) -- it's not as if it's
> evil, or even particularly tricky...
>
> -- chris

Agree with you Chris,

Java is perfect for this and it's not a hard thing to do, but the benefits might be tremendous. Java RMI has become a lot simpler with java 5 as the stubs and sceletons are now generated at runtime, not development time. What is the point of generating java classes and then noting that developers can't change them? If I can't change a class then don't show it to me. Or, what is the costs of having millions of java developers having to do an extra step in development to generate these things, and then maintain them (some will actually check them into version control systems)? The costs are huge. Java technology stack is surpassing human capacity to understand it, and it is overdue for simplification, IMHO. Received on Sun Aug 27 2006 - 17:46:07 CEST