[Info-vax] Reimplementing VMS, was: Re: HP adds OpenVMS Mature Product Support beyond the end of Standard Support

[johnwallace4 at yahoo.co.uk]

On Monday, 3 February 2014 09:17:38 UTC, Johnny Billquist  wrote:
> On 2014-02-02 19:34, JF Mezei wrote:
> 
> > On 14-02-02 18:08, Johnny Billquist wrote:
> 
> >
> 
> >> At no point is any code actually being written as a part of this. It's a
> 
> >> feasibility study.
> 
> >
> 
> > We'll never know. Such code is probably hidden very deep. It is quite
> 
> > likely proof of concept would have been done for very early stages of
> 
> > boot. Remember that VMS already knows about EFI so booting from EFI
> 
> > 8086s should be easier.
> 
> 
> 
> And the fact that various people on the inside repeatedly saying that 
> 
> nothing was done just don't do it for you, obviously.
> 
> 
> 
> > Knowing if existing X86 compilers can handle VMS code is an important
> 
> > part of an evaluation.
> 
> 
> 
> Which also have been said has not been done.
> 
> 
> 
> So, I repeat my request that you stop spreading unsubstantiated rumours 
> 
> that in fact have no other source than you.
> 
> 
> 
> >> This part sounds like just nonsense. What has things done in 2004 have
> 
> >> to do with anything?
> 
> >
> 
> > 2004 is when Intel relented and announced 64 bit 8086s.
> 
> > 2004 is when Intel announced CSI for both IA64 and 8086 (granting 8086
> 
> > servers same memory performance for multi processor/core systems as the
> 
> > "high perfrormance computing" IA64.
> 
> >
> 
> > 2004 is when HP announced IA64 would be reduced to HPC niche, dropped
> 
> > workstations and low end servers as a focus for growth.
> 
> >
> 
> > As a point of reference, Apple announced its move to Intel mid 2005.
> 
> > First 64bit Mac was in Aug 2006, full transition to 64 bit chips by 2007.
> 
> >
> 
> > 2007 is when the first life support deal for IA64 was signed, 2010 is
> 
> > when it was renewed/re-adjusted to extend the life again (spread the
> 
> > releases over longer period). (Remember that Tukwila, the first CSI (now
> 
> > Quickpath) based IA64 was originally due 2007.
> 
> >
> 
> > 2009 is when dismantlement of VMS engineering began.
> 
> >
> 
> > So any pilot project to look into porting VMS to 8086 would have been
> 
> > between 2004 and 2009, likely just before/around 2007.
> 
> 
> 
> That is just pure speculation on your part (the assumption on *when* 
> 
> something was done). And in addition, when a feasibility study can be 
> 
> done way before or later than that without this being the point at which 
> 
> the future on VMS would be decided.
> 
> 
> 
> A feasibility study gives you some data. Decisions are made based on 
> 
> data, and decisions can change. It has very little to do with *when* the 
> 
> study was made.
> 
> 
> 
> And you are once more trying to confuse the act of "look into porting" 
> 
> and "porting", which are two totally different things.
> 
> 
> 
> >> You can do a feasibility study at any point.
> 
> >
> 
> > Until Intel announced 64 bit 8086s a port of VMS to 32 bit 8086 did not
> 
> > make sense. And until Quickpath arrived for 8086 (circa late 2008 as I
> 
> > recall), multi-core/processor performance for 8086s did not scale that
> 
> > well in a large data centre.
> 
> 
> 
> While I suspect that there was not much point in porting to the 32-bit 
> 
> x86, any porting to 64-bit x86 would have little relation to multicore 
> 
> and multiprocessor issues, since those are changing all the time anyway.
> 
> 
> 
> >> The problem is when people start thinking that DEC/Compaq/HP wrote some
> 
> >> code to actually get VMS running in x86. Which does not exist.
> 
> >> Studies are a different thing than code.
> 
> >
> 
> > "writing" code. No. But test compiling with the 8086 compilers yes. They
> 
> > would need to evaluate differences in code generated and syntax
> 
> > translation to know how much code syntax/structures might have to be
> 
> > changed.  (and look into compiler/linker issues for off platform
> 
> > building, if any.)
> 
> 
> 
> You know. Of all the things you have written, this was probably among 
> 
> the silliest things so far. What 8086 compilers?
> 
> 
> 
> 	Johnny

"What 8086 compilers?"

Historical interest only: were you aware that there was a subset
Bliss compiler for x86(32)? It was reportedly used in the days when
Rdb for NT was considered a goer, but was very much cut down from the
real thing. There's been discussion of it in this newsgroup ages ago.

Also there was Compaq Visual Fortran for x86.

So, there's one and a half, maybe. One of them might even be relevant
to VMS the OS. So how many important compilers (and, obviously,
associated tools, runtimes etc) does that leave missing? "Lots", maybe?

Also, does the name Bill Todd mean anything to you? On 1 Feb 2006 Bill
Todd wrote here in comp.os.vms, in Niel Rieck's "Pentium Chronicles"
thread [1]:

"active investigation into porting VMS to 32-bit x86 *did* occur,
as part of the Emerald(?) project at DEC (IIRC something like 15
years ago).  How far it actually got I don't know"

And I have no idea how accurate that statement is, especially when
taken out of context like I just did.

John Reagan, on the other hand, I do take as definitive. He also has
a couple of posts in that thread, about GEM on x86, and what it could
(and mostly could not) do.

See e.g.
http://unix.derkeiler.com/Newsgroups/comp.os.vms/2006-01/msg01832.html 

Reminder: historical interest only.