On Jun 5, 2004, at 9:17 PM, Lex Spoon wrote:

>> The first step would be to create a binary format for distributing
>> Squeak code that would be faster to load than fileOuts - ie, 
>> decompress
>> the AST and generate byte code without having to do any parsing. 
>> Beyond
>> that would be modifying the interpreter to execute the binary format
>> directly.
>
> Note that there are differences between compressing for distribution 
> and
> compressing to save space in an image.  In the former case you need to
> go all the way down to bytes, but in the latter you can choose to use
> object pointers in some places.  One place this would likely help is in
> encoding symbols; you can store #nextPut: as a 4-byte object pointer
> instead of an 8+ byte string "n e x t P u t :".

Agreed. The paper I cited earlier mentions replacing strings in the AST 
with indexes into a string table, as the same strings tend to appear in 
several places in the tree. So the only real difference would be the 
table used to intern symbols - for distribution a local symbol table 
would be included in the file; within the image, we'd just use the 
regular symbol table.

I suppose there would be some processing required to load a slim binary 
method from a file into the equivalent of a CompiledMethod. References 
to shared variables would be resolved, literals converted to objects, 
symbols re-interned etc.

Colin