Re: Java performance on the i ? Good? bad? OK?

Buck wrote:

One further thought: I would never, ever, never think of developing Java
on i. That is, I would not try to use EDTF and JAVAC to edit and debug
my Java. I do it on a PC with WDSC. Yes, even testing. There are a
relatively small number of Java applications that MUST be run on i -
almost everything can be tested on a PC using TCP/IP connexions to i.

IMO: If you have an application that MUST be run on the i, then you've probably designed it wrong.

david

That's a darned good observation. The whole point of the VM is to insulate you from the underlying OS and IMO IBM's Java implementation on the i does that in spades. When I first found out that I could take a Java class and move it directly to the i and run it, it impressed me mightily. But then I did some more thinkin' on it, and I realized the machinations that were required to get a Unicode VM to work on an EBCDIC machine, and then I was blown away. And when I finally realized that I could just compile to the IFS and use a mapped drive to test from the PC, I finally realized what platform independence really meant.

Anyway, the few things that absolutely require running on the i (things like JNI calls) are usually unnecessary for all but the most sophisticated applications, and most of the time you can do all your work on your little PC and then load it up to the host for system testing.

Joe