On 11/21/06, Mike Heath <[EMAIL PROTECTED]> wrote:
On Fri, 2006-11-17 at 14:47 -0600, Greg Duffy wrote:
> Direct buffers are, AFAIK, the only kind that can be accessed with JNI,
> which I use extensively. Just my $0.02.
I agree. I got tired of waiting for JSR 203 so I have been
experimenting with the POSIX aio library in JNI. I currently have
asynchronous file reads working and my initial experiments indicate that
direct buffers are significantly faster than heap buffers when used with
JNI. Heap buffers can be used in JNI but you have to jump through hoops
to create a Java byte array and pass the byte array into the buffer or
vice-versa if you want to read the contents of a buffer.
Once/if my AIO framework stabilizes, I'll run some additional
experiments comparing performance between heap/direct buffers in
standalone mode and with MINA.
If anyone's interested, I would be happy to post my AIO code in my
sandbox once it's in a state that others might find useful.
It would be amazing to see an AIO framework in Java. The IBM guys
don't seem to be interested in open-sourcing AIO4J anytime soon/ever.
Do post up if you need help with the library.
Regards,
Vinod.
