[Openmcl-devel] Re: Trying to get LISA to run

Gary Byers gb at clozure.com
Tue Mar 29 13:51:52 UTC 2005



On Tue, 29 Mar 2005, nrms wrote:

> Okay, no wonder nobody replied, since what I wrote was nonsense.

Sorry. Replying was on my to-do list.  Honest.

> Actually there 2 problems, the first one is that require stuff (that
> one I wrongly thought caused all this mess). I guess this one I could
> circumvent by loading the needed file explicitly.

CCL:*MODULE-SEARCH-PATH* contains a list of directories; REQUIRE
looks in each of these for a loadable file that "matches" the name of the
module.  So,

(defun lisa-debugger ()
   (translate-logical-pathname "lisa:debugger;lisa-debugger.lisp"))

#+openmcl
(pushnew (pathname-directory (lisa-debugger))
         ccl:*module-search-path*
         :test #'equal)

should cause (

? (REQUIRE "LISA-DEBUGGER")

to find and load "lisa-debugger.lisp".

(There are other ways of controlling REQUIRE's behavior; the one based
on *MODULE-SEARCH-PATH* is built-in and pretty simple, but you can
define your own functions to map module names to one or more files.)

>
> The major problem seems to be that in a file called epilogue.lisp I got
> the following code
>
>   (set-dispatch-macro-character
>       #\# #\? #'(lambda (strm subchar arg)
>
> which is the evildoer. Is the problem here that character "?" is the
> prompt in openmcl?  Is there an intelligent way to solve this conflict?

OpenMCL uses #? to reference foreign variables, as in:

 ...
(if (zerop #?errno)
  (format t "~&No C errors lately."))

(That's a bad example, because "errno" may not be a real variable in a
multi-threaded environment.)

#? is used to reference some foreign variables in the Cocoa example;
redefining the reader macro would probably cause problems when loading
that example.


>
> Thank you for any hints.

It should probably be easier than it is to get your hands on a totally
"standard" readtable (one that only has CL functionality defined, and
doesn't include extensions like #_, #$, and #?) in OpenMCL.  The fact
that that doesn't happen that way is probably a bug, but it wouldn't
necessarily prevent conflicts like this if it was fixed.

I'm not really sure how LISA's used, but it seems that the available
options are:

1) Use a different dispatching macro that happens not to conflict with
the one provided by the implementation (e.g., #!, maybe.)

2) Use the same dispatching macro and clobber the implementation's
definition (making it harder to load examples or other code that uses
OpenMCL's #?.)

3) Use separate readtables (with separate definitions of #?).

(3) is probably the cleanest solution, but may be somewhat awkward for
users.

>
> Regards,
>
> Nils Stritzel
>
>
>
> Am 28.03.2005 um 18:18 schrieb nrms:
>
> > Hi!
> >
> > I am trying to get LISA (Lisp Intelligent Software Agents) to run with
> > openmcl. Actually I have already thought I had suceeded, but as it
> > turned out, I haven't.
> > The following is how it is supposed to work with other lisps. I tried
> > to figure out one solution myself, but it failed with the interesting
> > side-effect
> > the (require "cocoa") did not work anymore after calling my "solution".
> >
> > (defun lisa-debugger ()
> >     (translate-logical-pathname "lisa:debugger;lisa-debugger.lisp"))
> >
> > #+Allegro
> > (setf system:*require-search-list*
> >   (append system:*require-search-list*
> >           `(:newest ,(lisa-debugger))))
> >
> > #+clisp
> > (setf custom:*load-paths*
> >   (append custom:*load-paths* `(,(lisa-debugger))))
> >
> > #+LispWorks
> > (let ((Loadable-modules `(("lisa-debugger" . ,(lisa-debugger)))))
> >
> >   (defadvice (require lisa-require :around)
> >       (module-name &optional pathname)
> >     (let ((lisa-module
> >            (find module-name loadable-modules
> >                  :test #'string=
> >                  :key #'car)))
> >       (if (null lisa-module)
> >           (call-next-advice module-name pathname)
> >         (call-next-advice module-name (cdr lisa-module))))))
> >
> > So this what I thought that it might be a solution. As I figured out
> > by now it is obviously wrong; ccl:*module-search-path* is not what I
> > am looking for,
> > since it only contains directories and I try to append a file name.
> > But what is the correct way to solve this?
> >
> > #+Openmcl
> > (setf ccl:*module-search-path*
> >       (append ccl:*module-search-path*  `(,(lisa-debugger))))
> >
> > Any help or advice appreciated.
> >
> > Nils Stritzel
> >
>
> _______________________________________________
> Openmcl-devel mailing list
> Openmcl-devel at clozure.com
> http://clozure.com/mailman/listinfo/openmcl-devel
>
>



More information about the Openmcl-devel mailing list