[plt-scheme] Native code generation and immutable pairs

On Feb 10, 2006, at 11:31 AM, Will M. Farr wrote:

> On 10 Feb 2006, at 10:39 AM, Matthias Felleisen wrote:
>
>> Please take a very close look at the constraints that
>> bigloo modules impose on Scheme code. I am almost certain
>> that these results do not carry over to other Scheme
>> implementations.
>
> As someone who is reasonably familiar with Bigloo, I think the only 
> extra constraint that Bigloo modules place on the code (aside from 
> handling syntax expansions very poorly compared to PLT-style modules) 
> is that the module declaration allows for optional type declarations.  
> Other than that, the "closed-world" assumptions required for type 
> analysis are equivalent between the module systems.

It is the closed world assumption that I meant. It is for that reason 
that most other Scheme compiler papers report much less of a speed-up 
than Manuel. The person who posted appeared to be unaware of these 
essential differences. (BTW, my claims for Soft Scheme were for a 
closed world, too.)

To this day, few Scheme systems support modules and compilers in this 
way. For a large part of the past 20 years, people were proud to 
support Scheme as an interactive open system and therefore 
underreported compilation results in some sense.

-- Matthias




> Bigloo's module system is certainly less flexible than PLT's, 
> particularly with regard to the initial language assumption and the 
> renaming of bindings on export/import, but I don't think these 
> differences would be relevant to type analysis.  On the other hand, 
> I'm not an expert in language analysis, compilers, PLT or Bigloo, so 
> take this with a grain of salt.
>
> Will
>
> _________________________________________________
>  For list-related administrative tasks:
>  http://list.cs.brown.edu/mailman/listinfo/plt-scheme