Thanks, Dan. You are right, it's best to separate these issues.
Cheers,
-Rick
Daniel John Debrunner wrote:
Rick Hillegas wrote:
I'm afraid I was too elliptical in my first attempt to describe this
problem. Here's my next attempt:
[snip]
The point I am trying to make is this: It seemed to me that code cloning
was put forward as an alternative to the compatiblity checking in
David's first proposal. I don't think cloning solves this problem. It
merely pushes the problem around a bit. We still need run-time
compatibility checking.
I think you are descibing a different problem, communication between the
client and the server. In that case yes, version compatibility checking
is needed.
But what David is describing with the common code in separate packages
does not need versioning, as the code is self contained within its
domain of engine, client or network server.
You are describing the case where the common code from one domain
creates something that's consumed by common code in another domain (java
object in serialized form, drda protocol etc.).
That style would need versioning even if we didn't have common code.
Dan.
