On Dec 24, 2023, at 9:23 AM, Bill Gunshannon via
cctalk <cctalk(a)classiccmp.org> wrote:
I am back to playing with RSTS/E 10.1 again and have a couple questions
if there is still anyone around with experience.
First: Is there a way to change the allowed length for passwords?
The minimum length defaults to 6; the max is 6 for passwords that can be looked up (not
hashed) and 14 otherwise. The max is fixed. You can change the min by patching module
OVR, address ..MPWD to the value you want.
Second: Is there a way to make login take the
assigned name rather than
the x,x format for logins? I seem to remember using a system once that
did but I have no idea if it was legit or a local hack. Although I have
no problem using local hacks. :-)
No. There was a "named accounts" A/D project in the V8.0 era, and fragments of
that may have made it out the door in that version. It worked reasonably well in the
development systems at the time, as I recall. But it was never finished. If any of it
remained in the code, that was all removed by V9. Interestingly enough, it certainly
could have been done in V9 more easily than in V8, given the new file structure, but none
of that was ever undertaken that I know of.
FYA, another incomplete piece of work (this one in V9.0) was "installed files".
Those were files left permanently open, so you could access them without a directory
search. That wasn't all that interesting, but a variant of that operation would
install the file with an associated privilege mask. The idea was for "privileged
programs" to be selectively privileged, just as account are in V9 and later. I got
that working but we never did the followup work needed to sort out exactly what privs each
privileged program actually needs, so the old approach (all or nothing) continued to be in
use.
Need to get a system going and maybe even join
HECNET.
I really wish there was TCP/IP for RSTS.
I found one for V8.0, described as by "Stacken, Royal Institute Of Technology,
Stockholm, Sweden". I haven't tried to use it. A logical approach would be to
update it for V10.1 and hook it to the standard Ethernet drivers in that release. The
original comes with a DELUA driver, and a document describing it -- in Swedish. It's
clearly not the same API as the Ethernet drivers in RSTS, which have an extended set of
"I/O service" function codes for use by kernel components like DECnet or LAT.
So, minimally, porting that code would involve updating it to that API.
I forgot where I found this code. Johnny, do you know about it?
paul