On Wed, 11 Dec 1996, Bill Studenmund wrote:

> > There's probably a reason for that... I'd go as far as to say that Mach
> > is a cool research OS (and, that's what it is; a research OS), but
> > I'm hesitant to base production servers and workstations on it.
> 
> You may have good reason to be hesitant. I'm not sufficiently familiar
> with Mach to know the issues. But because of Apple's support of the
> MkLinux project, people ARE going to base "production" machines on
> it (or at least try to).

Judging by the way things are going, MkLinux is a fairly stable setup.
The only problems I've had with mine have been fixed in subsequent (DR2)
updates, but I haven't have time to download the 100 Meg DR2 package over
a slow modem link.  (I'll take my HD on campus over Christmas, perhaps.)

> I think a port of Lites would get us up &
> running quick, and give an alternative to Linux. And we let Apple
> port it to all the macs at first. :-)

I for one am still planning to at least keep looking at a NetBSD port to
mach, but if that fails, I'd love to have a port of Lites to fall back on.
I just pulled the Lites sources as a starting point for porting NetBSD,
will pull the Mach microkernel sources soon.  (I took my last final,
Pascal, this morning, so....)

I also talked (by email) to Nick Stephen, one of the people at OSF-RI,
responsible for the Mach3 port to the PowerMac, and he gave me some good
info on how to test Mach Servers without restarting, via a Mach-aware gdb.
He also says that no modifications are needed to the existing gcc for
MkLinux to compile servers for Mach3, unless you are collocating (which
will be done eventually, when enough of the bugs are worked out), in which
you have to specify where you want the binary to load (a change in the way
the binary is linked).

Incidentally, the Mach mk implementation on the PowerMacs was designed
with ufs (as per OSF/1) filesystem support in mind, along with others.
I'm assuming the ufs/ffs implementation used by NetBSD is the same, right?  
The Microkernel calls the bootstrap process with the partition info, and
the bootstrap process handles reading the partition, and is capable of
handling "various filesystem formats and executable formats", but that
the support isn't compiled in to the stuff for the PowerMac port, at
least not in the copies distributed.  But still, that's just a recompile
away, possibly without any coding needed for mk filesystem support at all.

> Though I still do like (and will help) a monolythic port too. I think
> both have advantages.

That would, indeed be a "Good Thing (tm)", as well.

Later,

 /---------------------------------------------------------------------\
|David A. Gatwood             And Richard Cory, one calm summer night,  |
|davagatw@mars              Went home and put a bullet through his head.|
|dgatwood@nyx.cs.du.edu              --Edwin Arlington Robinson         |
|http://mars.utm.edu/~davagatw -or- http://nox.cs.du.edu:8001/~dgatwood |
 \---------------------------------------------------------------------/

Requisite notice of "Good Thing" being a trademark of somebody else.