John M McIntosh wrote:

> 'could' because I'm not sure running a x-windows based Squeak is the 'best'
> solution for the mac. Doable, which is great, but is it acceptable? Mind of
> course Squeak has no mac interface so many of the issues are mute.
> 

> I think most of the Open Transport work I did will migrate to Cocoa.
> 
> Seems I'm forgetting the Cocoa work that has been done, my point was that
> you should be able to take the basic x-windows code and compile that and run
> it on OS-X, but as you've stated a true Cocoa application is better.

John,

MacOSX does not use X, nor even has it installed (but there is a port
available, done by John Carmack). This has always been the case also for
NextStep et al.

Secondly, I have thought for a long time that the various involved VM'ers
should sit down together (virtually perhaps) and agree on how to do the Mac
VM for OSX. It seems unwise to port Squeak's sockets-based VM code to the
streams-based Open Transport, only to move it to Carbon/OSX, where Carbon
merely interfaces to OSX's BSD sockets. Especially when there is already
Marcel's version that uses the built-in BSD sockets directly, without three
levels of indirection.

So far, I've had no success in convincing anyone about this. Great, we'll
have _two_ versions of the VM for OSX, one inheriting from Unix/NextStep,
and one from the classic MacOS!

Marcel, how's the performance of the Cocoa-based version on OSX, compared to
the Mac VM on OS9 (in other words, what would you guess it would be compared
to a Carbonized Mac VM running on OSX?)

Henrik