Re: State of play?

On Jul 16, 2008, at 1:55 AM, Peter wrote:

> I've been in a split world of real code in MCL and parallel threads
> of exploration in CCL for more than a year.  With news that Snow
> Leopard may not support PPC systems, I'm reminded that more G5
> hardware failure may make my lisp coding days on Mac problematic.
>
> What news of progress with MCL?  Am I hearing echoes in a dark
> chamber of legacy ... (I can see the shadow of my pension (Symbolics
> iron) far down in there already).

I'm back from vacation and have some (hopefully) good news to share on  
this subject.  Clozure has gotten a contract to port significant  
portions of MCL to CCL running on Intel Macintoshes.  We're currently  
working on this, and making good progress.  When it's further along,  
the resulting "MCL-in-CCL" will be released as open source.

Now, there are some caveats: the singular goal of the contract is to  
get our client's large gui-intensive MCL application running on Intel  
Macs with as few source code changes as possible.  That goal is  
driving our implementation work.  MCL-in-CCL will support the features  
needed by this application.  Because the application is large, it will  
as a matter of course sweep along with it a good percentage of MCL GUI/ 
Macintosh functionality, including windows, dialogs, menus, etc.  But  
it won't get everything, some things will work differently, and some  
things won't work at all.  In particular, events may be handled  
somewhat differently, threading is handled differently (CCL has real  
preemptive threads), some obscure features (e.g. colored menu items)  
may never even be tested, etc.  Most MCL source code will require some  
updating to run, though we do have the goal of minimizing that.

The big thing that is not included in this project is the MCL IDE.   
Now, the MCL IDE is written in MCL and is built on all the stuff  
described above, so it might not be too much additional work to get it  
running.  But we don't really know at this stage.  It will hopefully  
be clearer when the project is further along.  At that point the  
question of how to proceed will be one for the community to decide.   
We'd be happy to finish the work on a funded basis, or it could be a  
community-developed open source project.

Of course, if anyone wants to help fund this development effort to  
make it more complete sooner, we'd be happy to talk with you about that.

A bit of background on how MCL-in-CCL is being implemented:  Generally  
speaking, we are proceeding by resolving differences in the MCL and  
CCL system call (nee "trap") interfaces, and then porting the MCL GUI  
libraries for windows, menus, quickdraw, etc to the resulting version  
of CCL.  The work is currently being done in CCL PPC-32, but it will  
be moved to CCL Intel-32 as that matures in the coming months.  This  
approach has the advantage of supporting all of the Carbon libraries  
that MCL has been using for many years.  Of course, it also has the  
disadvantages that come with Carbon and 32-bit computing: they  
represent yesterday's APIs rather than tomorrows, they are not cross- 
platform, etc.

Still, we expect MCL-in-CCL to be a very useful way to provide life- 
extensions to many existing MCL applications and MCL programmers.   
That gives us time to continue to explore the other options people  
have raised for getting a longer-term more forward-looking solution.

I'm happy to answer questions about this as much as I can.   
Unfortunately, I'm not the only Clozure person who is taking summer  
vacations, so there are some questions that I may not be able to  
answer for a few weeks until we're all back.  But I'll do my best.

Andrew Shalit
Clozure Associates

> Finding the art of falling in love with Hemlock a tab counter
> intuitive (not helped by Clozure's being elsewhere occupied than on
> their IDE at present).  But that and diving into the Cocoa labyrinth
> seem the only visible glimmers at the moment.  Or get more serious
> about Lispworks.
>
> I've long felt that investment in pure CLOS and OpenGL might last a
> while before Apple and it's NeXT 6 shooter blow my investments out of
> the water.  But just now, even those anchors don't seem so safe
> (complexity and lack of robustness anywhere are not encouraging).
>
> Alex, how is Open Agent Engine going?
>
> Anyone care to review the state of play?
> Or has everyone gone off into the iPhone SDK sandbox!
> _______________________________________________
> info-mcl mailing list
> info-mcl@...
> http://clozure.com/mailman/listinfo/info-mcl

_______________________________________________
info-mcl mailing list
info-mcl@...
http://clozure.com/mailman/listinfo/info-mcl