Along the lines of the old java motto... "Write once, run anywhere"... those of us in the j2me development community all know that there are MANY inconsistencies between JVM implementations... and the problems that these cause for us all with having to write custom versions for specific devices. So here is my rant... (this actually comes from a reply I made on www.j2me.org)
I use folder names for each device type to hold the resources (images) and then set an ant property on build to the correct one and it includes the correct images...
In my opinion the big problem is more the non standard implementations and the inconsistencies that this brings:
- Some that DO call hideNotify() etc... some that don't
- Some that do have the ability to know when a sound has finished others that don't
- Some that do have the correct getHeight() getWidth() values, some that dont.
- etc...
Here's the one of the big problems in my opinion...
No list of known faults.. ok, some manufacturers do keep a list of some of them, Nokia for instance has their "known issue" documents, but what would be REALLY helpful would be something like a bugzilla for j2me implementations (which would include things like firmware revision details too !!!)
There HAVE been a few attempts at this in the past... the University of Dortmund had a web application which had some details, but its a bit out of date these days. Someone has started some sourceforge project, j2mebugs I think it was, which is along the right kind of idea. However, nothing exists "in the wild" for us to actually use, wouldn't it be EXTREMELY helpful if the manufacturers and jvm developers themselves actually let us know about things, even if it means just a press release or a known issues document that we can then include the details of in a database ?
I go back to my argument a while ago ... if the jvm vendors (and manufacturers) can't stick to a standard how do they expect the javaverified.com process to work for us ? The Technology Compatibility Kit from Sun, SHOULD test for these kind of problems and anyone who doesnt correctly meet the standard SHOULDNT be licensed... that said, of course there are sections of the specifications which obviously can be interpreted differently... but then ... that SHOULD be answered by the expert group for that JSR

Ok, you've read my opinion, now lets have yours ?