Re: Is guile byte-code compiled?

[Tanel Tammet <tammet@cs.chalmers.se>]

Christian Lynbech wrote:

> We have tried to apply hobbit to a larger part of the main code of our
> project, but unfortunately without success, since hobbit in its
> current state has too many compromises wrt. to full (guile) scheme.

As an original author of Hobbit, I am of course interested in making
Hobbit a standard tool for Guile. Obiously this means that Hobbit should
be able to eat all the Guile scheme. However, I keep hearing that it does
not. On the other hand, Hobbit does eat all the scm (report 4) scheme except
certain complicated kinds of MACRO extensions to report 4. 

I am an active user of Hobbit for my own theorem prover projects, but there
I use report 4 scheme and an old scm version 4 as a basis. I have never tried
to migrate to Guile. For 'standard scheme' and scm Hobbit is IMHO as good as
you can get speedwise - it is very close to compiling stuff in C (if you do not
rely on call/cc and continuations in speed-intensive parts).


Considering Guile: unfortunately I am very busy and I cannot put a 
reasonable amount of time into developing or supporting Hobbit for public use.

Fortunately, Hobbit is public domain and has a maintainer - Bernard Urban.

I guess that we should make a small coordinated effort to 

* first, make Hobbit eat all Guile scheme
* second, make the compilation & linking process simple for Guile
* finally, integrate Guile-Hobbit to the standard Guile distrubution

Bernard, what parts of Guile-Scheme does Hobbit refuse?

Anybody out there willing to put some effort into these three steps? 
I am available as a virtual consultant - if necessary, I can 
explain Hobbit innards, etc, but I cannot guarantee I will be available
for actual coding, not to speak of maintenance.


Regards,
         Tanel Tammet