Re: qcmu (was: Attaching to a running CMUCL process)

   > PS.  I was looking over the recent thread mentioned
in parentheses in
   > the subject line -- I too am interested in attaching
to running CMUCL
   > processes.  I'd be curious to know whether Denis
Papathanasiou was
   > satisfied by the lispserver info (if he's reading
this).

   In the end, I wound up using detachtty/attachtty for this
(i.e. invoke
   the process under detachtty, accessing the REPL using
attachtty when
   necessary) and it really works well.

My interest is in creating a MUD-like environment with
distinct users
interacting with the same LISP process, so I don't think
this approach
would work for me.  

On 2/16/07, Joe Corneli <jcorneliplanetmath.org> wrote:
> My interest is in creating a MUD-like environment with
distinct users
> interacting with the same LISP process, so I don't
think this approach
> would work for me.

You're right about that: if multiple users invoke attachtty
on the
same process, only one attachtty "client" (the
most recent one) can
get REPL commands to work.

There may be options in either detachtty/attachtty which can
get
around this, but there was nothing obvious in the man
pages.

If qcmu is no longer maintained, have you considered
revisiting the
original lispserver idea
(http://
www.cons.org/cmucl/doc/lispserver.html) and writing your
own
version of it?

The original article on cons.org plus the qcmu source in
combination
may provide enough information to get it done.

   If qcmu is no longer maintained, have you considered
revisiting the
   original lispserver idea
   (http://
www.cons.org/cmucl/doc/lispserver.html) and writing your
own
   version of it?

   The original article on cons.org plus the qcmu source in
combination
   may provide enough information to get it done.

This is a sensible thing to do, but it might be easier to
find some
free MUD out there, and use that to run the user-interaction
part of
the server, then just hook this MUD into LISP to run the
back-end.