[Info-vax] Orphaned processes on OpenVMS

[Johnny Billquist]

On 2011-05-23 06.08, Bob Koehler wrote:
> In article<ir740v$3lg$1 at Iltempo.Update.UU.SE>, Johnny Billquist<bqt at softjar.se>  writes:
>>
>> Yes, VMS is newer. But no, I don't think Unix had much, if any,
>> influence on design decisions in VMS. Maybe MULTICS did have some
>> effect, but I wouldn't bet on it. When VMS was bring developed, Unix was
>> still somewhat obscure, and only used to some extent in the academic
>> world, and still a bit primitive by todays standards. (We're talking
>> 1975 here.)
>
>     The folks who designed VMS can speak for themselves, but what I've
>     heard is that they were well aware of UNIX, it's featurs, and it's
>     limitations.

You are right, the people who designed VMS can speak for them self. 
However, I doubt any of them are around here. :-(

But you do know that in 1975, the PDP-11/70 was introduced? At that 
time, Unix still only ran on the PDP-11/40, PDP-11/45 and similar small 
machines? We're talking about something that most Unix afficionados 
today would barely recognize. It's capabilities and features were way 
short compared to something like what was envisioned by VMS engineering.
We're talking about Unix V6, which was released in spring 1975, and 
which was the *first* release to see a wider distribution outside of Bell.

I have no doubt that they knew something about Unix, but I wonder how 
much, and even more wonder what impact that could possibly have had on 
VMS at the time.

DEC themself did not start any Unix distribution until Unix V7 (which 
itself was released in 1979, by which time VMS was already released).

It is, however, easy to see RSX in there, and the obvious ways they 
extended on it (not surprising, considering that VMS was based on RSX). 
RSX normally allows only a two level structure of directories (however, 
it is easy to extend RSX to allow arbitrary depths). Going from two to 
some random large number was a very small step.

>     A lot of past work at also DEC influenced VMS, but some of that work was
>     also influenced by earlier work outside DEC such as UNIX.

My point was that Unix at that point was hardly spread anywhere, and it 
was still a very primitive system. But no doubt the engineering looked 
both inside and outside of DEC to get inspiration.

>     The only thing I've heard substantially influenced by UNIX, and also
>     attempted various ways in earlier DEC operating systems, was the
>     now ubiquitous nesting of directories inside directories in a file
>     hierarchy.  Even then, there were influences that lead VMS to a
>     per-disk hierarchy, rather than UNIX style mount points.

All DEC OSes (that I know of) have had per-disk hierarchies. Nesting 
directories inside directories is an old concept, for which Unix can't 
really take any credit. RSX do the same thing, and I'm sure lots of more 
examples can be brought out.

>     So VMS users can be glad that user names and directory names are
>     strings, that subdirectories are part of the file system, but were not
>     done in the TOPS-10 nor TOPS-20 style.

What do you mean "strings"? Directory names in VMS are normal file 
names. Until VMS V4, this meant the same things as in RSX, ie. 9+3 
character filenames with only radix 50 character set used.

TOPS-20 style filenames and directories were much more free form, but as 
far as I remember, usernames and directories were sortof related, and 
kept separate from the file system. But they are certainly strings.

	Johnny