[l4-lisp] Current state: Beginning prototypes

[Julian Stecklina]

Hello,

just to get something onto this list, I will describe the current
state and plans of the project.

The goal of this mailing list is the implementation of what is now
"codenamed" VIER, which will be an implementation of a near-Common-Lisp
on top of the L4 microkernel. Design goals are reasonable efficiency in
standard cases and suitability for real-time applications. We hope that
VIER will be a noteworthy alternative to C/C++ as implementation
language for L4 servers. The best case would be that VIER servers are
indistinguishable from C servers to the rest of the system.

Of paramount importance will be the ability to cross-compile code from
existing UNIX-like systems (GNU/Linux, *BSD). Therefore the
implementation language of VIER's compiler will be (a subset of) Common
Lisp itself. We anticipate that the compiler itself will not be
available at runtime on L4 until the project has advanced to a
sufficient state.

At the moment we prepare a prototype compiler that will enable us to
play with various aspects of the implementation itself. My part will be
the selection and implementation of a suitable garbage collection
algorithm. Candidates at the moment are a Baker-style copying collector
with various changes to make it efficient on a virtual memory system
and a real-time-enhanced version of mark&sweep.

Current problems are the general lack of useful information about
memory management on L4 X.2 itself. Any help in this and other areas is
appreciated. Pointers to code will be posted in the near future.

Regards,
-- 
Julian Stecklina

"We were not out to win over the Lisp programmers; we were after the
C++ programmers. We managed to drag a lot of them about halfway to
Lisp." - Guy Steele, Java spec co-author