I recently learned that there is already a gcc cross-compiler for the Palm
Pilot.  "ASDK - Alternative Software Developer's Kit"  So that would be the
starting point for a GNAT port.

Also, someone said "the Java compiler for the Palm Pilot generates native
code, not J-code." This is not entirely correct.  I have not been able to
find a JVM for the Palm Pilot, NOR a "Java compiler for the Palm Pilot."
But I did find a program that translates J-code (not Java) into Dragonball
assembly code and the assembles it into a .PRC file.  "JUMP" was the acronym.

If anyone is interested (and can't find it in a Web search) write me and I'll dig through my filing NON-system looking for the URLs.

> From back in April:
>
>    Which Motorola chip?  If 680x0, it shouldn't be hard to make a cross-compiler
>    for PalmOS from some other GNAT 680x0 cross.
>
> It's their so-called "Dragonball" CPU.  I think it's designation is something
> like MC68342 or the like.
>
> I asked this very question on gnat-chat a couple of months ago.  After some
> discussion, the following points emerged:
>
>  - Yes, such a port was feasible.  All it would take is time, patience, and a
>    modicum of experience making GNAT or at least GCC cross-compilers.  Since
>    GCC already targets a very similar variant of the CPU (the 68320,
>    perhaps?) little if any changes to the code generator would be required.
>
>  - Ada tasking wouldn't be supported, but that was deemed not to be a problem,
>    since PalmOS provides all the application switching and background
>    execution support one would need.
>
>  - In order for the port to be useful, one would need to develop a binding for
>    the PalmOS API.  Not a trivial task, but not rocket science, either.

ASDK has the C binding.  Run the c2ada binding generator.

> I keep hoping some enterprising soul will come along and just do it, so I
> don't have to. <grin>  But if no one has by the time I get around to it (don't
> hold your breath) then I'll tackle it myself.