Misc ideas & comments

Rodrigo Ventura yoda@isr.ist.utl.pt
Fri, 27 Mar 1998 14:56:12 +0100 (GMT+0100)

        1. I tried RScheme, but it seemed to me too complicated, at
first in a first glance. After compiling everything, the whole tree
ocupied about 70MB of disk space. I had some difficulty to find the
main executable in the dense tree structure. I had to use "find
. -perm 111 -type f" to find it! But of course, this is my ignorance
speaking... The bottom line is that RScheme might be too complex to
allow a fairly good degree of exploration. And by the way, does
RScheme uses a virtual machine?

        2. Portability is a nice feature for LispOS. The easier way to
accomplish this is to make a platform-independent virtual machine. Of
course there are other paths, but they seem too complex to me
(gcc-like compiler).

        3. Is the gcc's RTL processor-independent? If so, how about
using the RTL-down part of gcc to make LispOS? What is the greatest
difficulty of compiling scheme? The macro scheme? The need to stuff an
eval function with the compiled code?

        4. The experience I made with scheme48 ontop linux used a
statically linked scheme48vm. But maybe we could use dyn-load
explicitly. How about considering the dyn-linking mechanism a part of
the Linux kernel? It could make things easier for LispOS. Or maybe
not, because scheme linking is probably more demanding than plain
C-style linking. Maybe we'll need to use a lisp-prepared dyn-loader.

        5. I think that debates like CL versus Scheme are useless if
they make the project stop. I don't mind switching to CL if that
proves to be easier and more efficient to get LispOS started.



*** Rodrigo Martins de Matos Ventura, alias <Yoda>
***  yoda@isr.ist.utl.pt, http://www.isr.ist.utl.pt/~yoda
***   Instituto de Sistemas e Robotica, Polo de Lisboa
***    Instituto Superior Tecnico, Lisboa, Portugal
***     PGP Public Key available on my homepage
*** Key fingerprint = 0C 0A 25 58 46 CF 14 99  CF 9C AF 9E 10 02 BB 2A