[Info-vax] Reimplementing VMS, was: Re: HP adds OpenVMS Mature Product Support beyond the end of Standard Support
JohnF
john at please.see.sig.for.email.com
Tue Feb 4 07:49:26 EST 2014
David Froble <davef at tsoft-inc.com> wrote:
> I've had conversations with people who would know about
> some of the issues with a port. It's not deep dark secrets.
> Not that any of this matters.
> 1) I doubt the feasibility of re-implementing VMS.
> 2) If (and that's still a big if) HP were to turn
> VMS loose, then a port would still be a big job.
> Note that all of VMS, including the layered products,
> and most importantly the compilers, would be needed.
Do you know if a re-implementation of vms system services,
rtl, etc, has been done, using unix ipc mechanisms, etc?
That is, I have various C programs originally written for vms,
which I'd consider porting if the effort were feasible.
On those jobs I have tackled, I've done it in two phases.
First, edit the code to make it "syntactically portable".
That is, remove descrip.h and all descriptors, ssdef.h and
all sys$() calls, etc. To that end, I wrote a little wrapper
library http://www.forkosh.com/sources/vmslib.c.html
whereby all these kinds of calls are replaced by "ansi standard"
calls using null-terminated strings rather than string
descriptors, etc. Of course, that wrapper library just
turns strings into descriptors, etc, and then issues all
the standard vms calls.
The second phase is "semantic portability", i.e., rewriting
that library using portable (by which I mean unix) mechanisms.
I've done that for just a few things like logical name services
(actually, just a small subset of that, whereby each table
is represented by an ascii file containing "name:translation;"
pairs). But a much more complete re-implementation is needed.
That would at least allow some vms application code to be
ported in a pretty mechanical way, i.e., just straightforward
editing to replace the vms-isms with functionally equivalent
calls. No re-designing necessary.
Anything along these lines already exist?
--
John Forkosh ( mailto: j at f.com where j=john and f=forkosh )
More information about the Info-vax
mailing list