Re: Code changes fixed problem with IBM JRE + .NET interop demo

Hi Roger, Robert and others

Roger I Martin PhD wrote:

> Cool!  I'll do some tests.  The things I tried over the weekend didn't go
> smoothly( parsing for non COM dll's).

First I am sorry that I don't have much to contribute on that issue - I
am not that well into the DLL entry and header-files stuff. So I must
admit that a lot of the discussions you, Robert and Cyril have had, went
well over my head ;-).

> But I would like to look at the
> refactoring of marshaling. What are your current thoughts on it?

There is of course the Java.NIO things we have been talking about
before. If we decide that we will require JDK1.4 (I am for), then it
would be great to try out NIO. I have not used that package yet (I am
still stuck with some JDK1.3 apps on my work), but I assume what you and
Robert have been discussing, must have been using java.nio.ByteBuffer
instead of the bytearray + NakedByteStream we use today? One way to do
the refactoring would be to extend the rough performance tests I have
started on (try to run the 'test'-target, and look in the
dist/test-reports/index.html - the org.jawin.perf-tests are actually
some kind of performance tests), and then run these tests before and
after refactoring to see what the gain of NIO in this context is.

Another marshaling refactoring/updating that could take place, would be
to go through all the other marshalling instructions than 'v' and 'V' (I
have refactored those), and make sure the java and native side are in
sync, and then make documentation for each instruction - I actually
hoped to get started on this within to long.

Best Regards
Morten