[Info-vax] Anyone interested in another public access system

[David J Dachtera]

Bill Gunshannon wrote:
> 
> In article <BZhjsJkqzq4v at spock.koehler.athome.net>,
>         koehler at eisner.nospam.encompasserve.org writes:
> > In article <72pgbbFrckbiU1 at mid.individual.net>, billg999 at cs.uofs.edu (Bill Gunshannon) writes:
> >>
> >> U*mmmm....  No, not really.  Active development (well, at least as active as
> >> VMS seems to be) for three of the PDP-11 OSes just ended last year.  So, if
> >> you take when they first started and compare that to when VMS first started
> >> VMS will need a few more real good years to match them.
> >
> >    Hm.  Although possibly not as important as the Berkley port to VAXen,
> >    I seem to recall the development of UNIX including a port to PDP-11
> >    early on.
> >
> >    Sadly, I think continued development of that '60s technology will
> >    continue for a bit longer.
> 
> Unix is no more 60's technology than VMS (which traces it's roots back
> to RSX on the PDP-11 as I recall) or zOS which can trace it's roots all
> the way back to the IBM 360.
> 
> It never ceases to amaze me that Unix, which has seen constant development
> and considerably more attention from Computer Scientists and Engineers
> than VMS is seen as stagnant while VMS which, as frequently stated here,
> has seen little in the way of anything beyond the minimal bug fixing, is
> seen as the epitome of modern computing.

Well, UN*X, at it's core, has changed relatively little since its
inception. Sure, there have been improvements in scheduling, memory
management, loadable kernel modules/drivers and the like, and in the IP
stack, the shells have acquired some features they lacked which
essentially bring them up to where VMS+DCL was 20 years ago, ...

... but, the whole system is still spawned from a single process (init,
PID 1) (VMS "fixed" that (sort of) over a decade ago (most processes are
still spawned by JBC, but the resulting process trees are not owned by
the JBC), VMS-style clustering is still non-existant, VMS-style
"cluster" management (ala SYSMAN) is still a site-specific collection of
cobbled-together kludges, the manpages are still terse, esoterically
worded and about as user-unfriendly as such a feature can get and still
exist, command names are still cryptic and non-sensical, command options
are equally cryptic, non-sensical and disuniform for the most part (IBM
has made some strides in their own add-ins, but zero elsewhere), the
filesystem is still fragile and easily broken (lose a directory with no
record of which inode number belongs to which file, and you're
hopelessly screwed), no RMS or equivalent, no logical names, multi-path
FC support remains a third-party add-in (example: EMC PowerPath),
privilege is still an all-or-nothing proposition, ...

Need I go on?

I've no way to know whether there is any hope of ever getting any new
blood in OpenVMS engineering, but I'm hoping someone, somewhere, perhaps
in VMS V10.0-1 will solve the "fork()" problem and in doing so solve
many of the incompatibilities between UN*X and VMS, perhaps even merge
UN*X and VMS into something that brings the best of both worlds to EDP. 

I'll be a worm feast by then, to be sure, but I hope it does someday
come to pass. I'll send a big, fat "I told you so!" from the other side
by what ever means may be available.

D.J.D.