🚀 go-pugleaf

In article <adtDKo622.Mtu@netcom.com>, adt@netcom.com (Anthony D.
Tribelli) wrote:

] Christopher C. Wood (christw@meaddata.com) wrote:
] : Yes.  Copland is based on a Mach microkernel.
] :
] : |> I thought threads have all sorts of restrictions as to what they
] : |> can call. Only one application thread can call quickdraw at a time,
] : |> similar for parts of the toolbox. Sounds like parts of copland are
] : |> not reentrant.
] :
] : A reentrant rewrite of the existing quickdraw toolbox is not part of
] : Copland.
] 
] Like I said originally, Copland is a Win95-ish upgrade. Win95 has new
] 32-bit pre-emptive multitasking but old 16-bit graphics code (among other
] things) that handicaps the system. Copland is in a similar situation since
] it is using old code also. In Copland, the Mach microkernel does not
] realize its full potential as it does in other OSs.
] 

This is simply not true, according to the information released so far
about Copland by Apple.  Copland will indeed by a micro-kernel based OS,
with fully-preemptive and protected processes.  One of these processes
(currently referred to as 'The Blue Box' will act as an environment within
which all 'old-style' Mac software apps as well as the user-interface
toolbox environment will run.  Thus old apps and the interface, while not
protected from each other any more (or less) than they are today in Sys
7.5.1) nonetheless are insulated from the kernel and other Copland
processes such as i/o and tasks that do not require a user-interface.  All
other Copland processes are protected from old apps running in the Blue
Box.

This is completely analogous to Win NT on a non-x86 machine in which old
16-bit (non-Win32) code is run cooperatively together as one NT process. 
Other NT processes are protected from this old 16 bit code.

Win 95 however currently allows 16 bit code to reach the lowest level of
the OS and hence any win 32 process in Win 95 is still vulnerable to old
16 bit code bring down the system.  The fact is, Win 95 itself contains
still a lot of 16 bit code.  Win 95 is very different from the way Copland
will be.

Note that the Copland and NT models for supporting old software are both
similar to the A/UX and MAE process model in which all Mac programs run
cooperatively inside a single unix process.  A really killer system that a
lot of us WISH would get released would be MAE running on AIX on PowerPC,
in which MAE would have the mixed-moded manager implemented so that native
PowerMac apps could run in it.  I have no idea why Apple has not released
an AIX version of MAE yet.  Pity.

The only place Copland fails to meet the process model of an OS such as NT
is that even newer, Copland - ready, applications will run the
user-interface part of the application in the Blue Box.  Hence, the
Copland model of app design will likely evolve into separating the
interface code from the core application so that the latter can run as a
separate protected process.  This is analogous to a current X application
where the X-client runs as one process that communicates with an X-server
for display.  In the typical X environment, bringing down the X window
server will bring down the display of all the X clients it is serving so
this is pretty analagous to how Copland will work.  However, most
interface code is fairly bullet proof, so I don't think that making all
apps put their interface in the Blue Box will make Copland all that
vulnerable, any more than X is today.

Cheers,

Dr. Mel Martinez
The Johns Hopkins University
Dept. of Physics
mem@jhu.edu