[: Re: ref: SISC for J2ME]

----- Forwarded message from  -----

To: eleftherios stavridis <[email protected]>
Subject: Re: ref: SISC for J2ME

On Sun, Jan 04, 2004 at 02:08:28PM +0000, eleftherios stavridis wrote:
> Hi Scott,
> 
> I assume that you are aware of the Java 2 Micro Edition (J2ME)
> platform[1], aimed at devices with limited memory and processing power
> such as mobile phones & PDAs. I am wondering whether it would be feasible
> to implement something like SISC for this platform (namely, for CLDC
> 1.0[2]/CLDC 2.0[3] & MIDP 1.0[4]/MIDP 2.0[5] class of devices) in such a
> way that complete applications can be developed in Scheme rather than in
> Java.

> 
> I will avoid to get into implementation details (such as the lack of
> java.lang.reflection) as I am only trying to put things in perspective
> for you. I would really like to hear your views on the matter, given
> your experience with SISC and last, to express my will to contribute
> a couple of devices for development purposes and $2000 USD upon (loosely)
> agreed deliverables (given that the code will be GPLed).

The idea is an interesting one, but from my knowledge of the MIDP and 
CLDC specifications, the biggest problem is available memory.  
SISC lite itself occupies nearly 200k compressed in storage.  There are 
places where memory usage could be trimmed, but it would be very 
difficult indeed to squeeze into the 32k + 8k profiles of those 
specifications.  The problem is largely that SISC is an interpreter, and 
therefor requires its engine and at least the runtime library used by 
the application that would reside on the device.  Obviously having done 
no research, its difficult to say how small this could be made.

A better approach would be to use Kawa or Bigloo, both Scheme -> 
bytecode compilers.  That way you would avoid the interpreter memory 
overhead.  Unfortunately you would not be able to do on device 
development with that mechanism.  

If you don't mind, I'd like to forward your request and this response to 
the SISC development mailing list, to get some other input.  I'll keep 
thinking about it as well.

	Scott



----- End forwarded message -----

--

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/+JSbr9IW4v3mHtQRAm/pAJ9AjzboTpSKpr3HxXDx6Al20rOj1gCbBfWO
56kAywkGGsR3yLn8dU6tEDc=
=rfSJ
-----END PGP SIGNATURE-----