Producing a different byte code set

Mats Nygren nygren at sics.se
Wed Sep 27 10:55:24 UTC 2000


Stephane,

Stephane Ducasse <ducasse at iam.unibe.ch> wrote:
> Hi
> 
> we are evaluating how difficult it would be to connect Squeak and  
> a mindstorm chip.
> A way to do it is to generate opcode for the firmware that is running
> on the chip. 
> So we were dreaming to change the byte code generated to generate e
> the firmware opcode of mindstrom. Then charge the chip via the right port. 
> 
> Apparently the byte-code seems to be hardcoded into MessageNode
> or one of its superclass  classVariable (I do not have Squeak open).

It's in ParseNode which is not the best place.

> Do you know if there is an object responsible for the bytecode set or if this 
> information is spread into all the parse tree nodes?
> 
> Have you any hints how to attack the problem besides reading the 
> blue book and the collective book chapters ;)? 

If Ian Piumarta is interested, that wouldn't slow you down ..
Marcel Weiher have done things ..

I'm working on a related thing. Refactoring the compiler so that
this should be both possible and easy. There are some postings in
the archive. I'm not going to study that processor (or others) in
the nearest future however.

At present I have to give priority to getting my masters thesis
finished, which is about syntactic versions of programs not bytecodes,
and I can give no clear timetable for the compiler refactoring.

The result of this when fairly finished should be a family of classes
that together makes generation of diverse material possible, different
byte-codes, parse-trees for other programming languages, pretty printed
versions, graphical, etc.

I have also though about how to handle codes in a better way within
Squeak-like systems. See "The Mosner bit" if you are interested.

/Mats





More information about the Squeak-dev mailing list