Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!apple!agate!ziploc!eps
From: eps@toaster.SFSU.EDU (Eric P. Scott)
Newsgroups: comp.sys.next
Subject: Re: Mac is software - why not on NeXT?
Message-ID: <1015@toaster.SFSU.EDU>
Date: 13 Nov 90 07:50:29 GMT
References: <1990Nov6.114141.3280@csc.anu.oz.au> <1990Nov6.160855.864@vax5.cit.cornell.edu> <1990Nov13.045946.13088@ux1.cso.uiuc.edu>
Reply-To: eps@cs.SFSU.EDU (Eric P. Scott)
Organization: San Francisco State University
Lines: 24

There's been a lot of "wouldn't it be nice if..." but is anyone
actually TRYING to write this?

The "it shouldn't be that hard" rationalization:

Both machines use the same architecture.  You should be able to
drop Mac code in a Mach task and expect it to run until it wants
something.

When UNIX code wants something, it executes a 680x0 TRAP
instruction.  When Mac code wants something, it executes
an "A-line" unimplemented instruction.  A UNIX process
that does this gets a SIGEMT signal.  Great!  Just write
a SIGEMT handler that looks at its return address and
does The Right Thing with the registers.


I tried calling a procedure with an "A-line" instruction
in it--with
	handle 7 nostop noprint
--and gdb still wouldn't transfer control to my SIGEMT handler.
Grr...

					-=EPS=-