Make LispM code FREE (fwd)

[Mike McDonald]

>From: eric dahlman <dahlman@cs.colostate.edu>
>Date: Tue, 31 Mar 1998 13:36:55 -0700 (MST)
>To: lispos@math.gatech.edu
>Subject: Re: Make LispM code FREE (fwd)

>As far as I can tell the sheet protocols and the different issues
>involving grafts and the such fall under Silica.  

  That was kind of my understanding too. But the CLIM spec doesn't
call them out separately as "Silica".

>I have the paper on
>Silica by Rao and my impression was that these protocols in the CLIM
>spec were Silica.  I could be wrong it wouldn't be a first.  The
>benefit is that if this portion of the system is well isolated it
>would be possible to take the CLIM implementation for say CLX and
>remove the CLX dependencies and use it to implement a version for say
>Windows.  Then code written for one system would also work seamlessly
>on another. 

  The protocols defined in CLIM are pretty well designed to separate
out the native windowing system dependancies from the core
protocols. In my implementation, there's one file called clx-bindings
that has all of the window system dependant code in it. This includes
the CLX specific classes CLX-PORT, CLX-GRAFT, and CLX-MEDIUM. The core
protocols are defined such that the right kind of thing gets created
when one is needed. (Usually by asking the port to make one.)

> And as far as the Lisp OS effort is concerned we would
>have a better chance for involving outside efforts in the effort.  I
>for one have not made up my mind as to whether I could sell everyone
>on having a Lisp OS computer on their desk.  However, if I could do all 
>my development on a LispM and then ship the resulting ap "Lobotomized
>for Windows End Users(TM)" I would be a happy camper. 

  I'm trying real hard to keep it from being CLX dependant. I'd also
like to see a Motif backend, in addition to a Windoze version.

>
>When you have something running let me know and I'll have a look at
>it. I would be very willing to put some time towards these parts of
>the system.  I may be a bit naive about this whole thing but looking
>at the CLIM spec there seem to be several large orthogonal protocols
>that could be implemented by different people in relative isolation
>once the system's basic framework is in place.

  That's precisely my plan. Someone could spend their life just
working on refining the presentation part,let alone the other
parts. I'd rather be using it! :-)

>It would be really nice if everyone could have the experience of
>working with a Dynamic Listener like on the Symbolics machines since
>this would make a lot of these issues clearer.  The system you are
>describing with presentation types and command tables would more than
>satisfy the shell crowd.

  YUP! :-)

>  And it would work fine on a textual basis.

  An ASCII terminal is harder than I want to think about at the
moment. I think it's certainly doable though, to some extent
anyway. (with-room-for-graphics might be a bit hard!)

>One possible glimpse of kind of what we are talking about is the
>"presenting listener" that Franz is including with the latest Emacs
>Lisp interface beta.  It is still buggy for me under Xemacs 20.4
>Solaris but it does have the ability to "present" objects to an emacs
>buffer and later "accept" them back.

  Heck! And here I am making the 4.3 final version work with CMUCL! I
bet they went and changed everythign on me! :-)

  Mike McDonald
  mikemac@mikemac.com