[slime-devel] wrapping a non-LISP prompt for use by SLIME

[Joe Corneli]

Hi,

I have been contributing some work to the Monster Mountain project,
http://code.google.com/p/mmtn/ -- which is nominally a Lisp-based MUD,
but is, more generally, a Lisp-based application server.  One of the
things that it currently features is a "multi-user Lisp", with login
via netcat.

In particular, it largely solves a problem I had brought up on this
mailing list in 2006 --

http://common-lisp.net/pipermail/slime-devel/2006-June/005033.html

I am now curious about what I think should be a rather simple problem,
which is, providing a SLIME connection to this prompt.

Schematically, what I envision is:

   remote server -runs-> LISP 1
   LISP 1 -runs-> Monster Mountain

   remote server -runs-> LISP 2
   LISP 2 -runs-> SWANK

   local Emacs -`slime-connect's over SSH tunnel to-> SWANK
   SWANK-embedded LISP process -logs in to-> Monster Mountain

   Result: SLIME interface to Monster Mountain.

This looks complicated, but my sense is that it should "just work" if
Monster Mountain can provide the "SWANK-embedded LISP process" with
the right stuff.  

I will probably have to ask some additional questions (probably not
here) to figure out how to get LISP to connect to Monster Mountain (as
above, currently that is done in the shell with netcat), but for this
list my question is: what should the interface of the Monster Mountain
server look like to its LISP client in order for SLIME to be happy?

(My intuition is that it should roughly look like we are in a "virtual
package", whatever that would mean.)