[no subject]

[Scott.Fahlman@SEF-PMAX.SLISP.CS.CMU.EDU]

Thomas,

Some of your questions will have to wait until William Lott, our Unix
wizard, returns from vacation.  That will be in a couple of days.  I'll
forward your questions to him.

In general, it is best to address technical questions to
"slisp-group@cs.cmu.edu" (plus whatever additional address magic you need
to reach the U.S.).  Such messages will be seen by our whole group and the
best person will respond.  Of course, anything of an administrative or
sensitive nature can still be sent direct to me.

I don't know what is in the kernel.core file you mention, but it's too
small to be a working Lisp core.  In any case, our core files will have
some built-in system calls that are Mach-specific.  It shouldn't be too
hard to port the system to another MIPS processor, but at the very least
you will have to replace some system calls and build a new core file from
scratch.  (It's possible that when the Mach-based OSF/1 system is widely
available, we will be able to run on that with no changes from one machine
to another, assuming all have the same CPU.  But right now, Unix system
vary a lot.)

I don't think our system uses threads in any significant way, but this may
be used in interrupt handling under Mach.  William can tall you more.

The new CMU Common Lisp no longer uses link tables.  I believe that both
stack-overflows and triggering of GC is handled by a guard-page scheme:
above a certain limit, the pages are protected, and Lisp gets an interrupt
when it hits these pages.  In versions of unix with less flexible
protection schemes, you may have to check explicitly whether things have
grown too large or risk disaster if something grows without bound.

  - Which scope do special declarations have (CLtL1 or CLtL2) ?

I'm not sure what you mean by this.  I don't recall any changes in the
scoping rules for specials.

  - Which MACH version is required by CMU LISP ?
    Are there major differences to our RT-version 2.5 ?

Currently, we require the very latest version of Mach for the Pmax.  This
is because some bugs were found in earlier versions.  The new Lisp is not
yet running on the RT, so I can't really say whether older versions of Mach
on that machine will be OK.

-- Scott