[Info-vax] Running OpenVMS native on x86 . . .

[BillPedersen]

On Dec 9, 11:43 am, Paul Sture <p... at sture.ch> wrote:
> On Fri, 09 Dec 2011 08:18:52 -0600, Bob Koehler wrote:
> > In article <R4CdnbG1qNAnQ33TnZ2dnUVZ_rWdn... at earthlink.com>, "John
> > Reagan" <johnrrea... at earthlink.net> writes:
>
> >> "Bob Koehler" <koeh... at eisner.nospam.encompasserve.org> wrote in
> >> messagenews:oBoFU6UNeH6E at eisner.encompasserve.org...
> >>> In article <DqSdnemgG-cx8UPTnZ2dnUVZ_u-dn... at earthlink.com>, "John
> >>> Reagan" <johnrrea... at earthlink.net> writes:
>
> >>>> Neither are from HP & OpenVMS Engineering.
>
> >>>   I believe Emerald was within DEC.
>
> >> Emerald was just a code name for some OpenVMS release (I forget which
> >> number it became).
>
> >    That may be true, but the Emerald project that I'm referencing was a
> >    bit of VMS on top of the Mach micro-kernel, on an 80386.  It got as
> >    far as first boot (login and do a COPY command).
>
> Type, edit and diff as seen here:
>
> http://www.sture.ch/images/VMS-on_Mach-Prototype-Session.png
>
> which is taken from page 13 of
>
> http://www.sture.ch/vms/Usenix_VMS-on-Mach.pdf
>
> --
> Fog differs from other clouds only in that fog touches the surface of
> the Earth.
>
> Paul Sture

But this shows that they still did the prototype on a VAX and does not
discuss any future work:

"Early in the project we decided that we would use a VAX platform to
produce the prototype. We chose the VAXstation 3100-48 system after
discussions with the CMU Mach team, who provided us with a version of
Mach 3.0 on that system"

Which is from the first paragraph of Section 3.1.

Now, do not get me wrong, I think this makes great sense as far as an
effort.  But the real issue of any port to any platform is whether you
will get the VARs and ISVs to follow.  OpenVMS does not have a good
track record here.  Each time it has been ported it has lost 30% or
more of its applications.

To suggest being able to get it on the X86 (X64) platforms will solve
this problem is most likely wishful thinking.  More effort has to be
given to compatibility and interoperability with the rest of the
players in this arena.

Porting Unix application to OpenVMS - some of the most interesting and
useful - is near impossible due to the lack of a real SSIO and a real
FORK mechanism at present.  This has been know for years and years.
It has been called the linch pin of getting Unix portability and it
still has not been completed.  Without these features even if it were
on the "industry standard" platforms it would still not be attractive
to port application to OpenVMS.

Bill.