1 Jun
2023
1 Jun
'23
8:55 p.m.
On Jun 1, 2023, at 2:39 PM, Thomas DeBellis
<tommytimesharing(a)gmail.com> wrote:
The most recent that I have from you for OST is 19, Ultrix-11. I don't have anything
past 9, OS-8 (!!). So I'll be delighted to shovel those into my sources, THANKS!!
I did have a few questions (naturally)
First, what do you do after you get the foreign configuration message in NFT? Send an
access complete and close or just close? I punt the connection, and was wondering whether
that might be considered a rude thing to do.
Rude or not, it's clearly possible: in any distributed protocol the connection can end
at any time.
Second, what does DTS/MVS mean?
Finally, Ultrix-32 as a file system reminds me of something that I had been thinking
about, which is that a number of operating systems can actually concurrently run multiple
file systems.
As an arcane example, one time I was walking past 2102 (Tops-20 Monitor Development) and
almost fell out of my sneakers seeing that it had a Tops-10 disk pack mounted. After
checking that it wasn't a dual port set exclusive to one of the nearby (beautiful)
KI's, I saw that Tops-20 was really using it. Wow. It's true, the hooks to use
Tops-10 file systems remain in PA1050. Tops-20 also treats DECtape as a different file
system, but I can't imagine what it would be reported as.
Clearly Unix can run a bunch of different file systems, so I won't get into listing
those.
Windows/NT, 2000, XP and I forget what else can run FAT12, FAT16, FAT32, HPFS and NTFS.
I had been wondering how DAP might accommodate this, or whether it needed to.
Those fields should be considered "for entertainment only". One serious error
made by certain implementations -- VMS was a notorious offender -- is to look at OS or
similar implementation info fields and make decisions like "if I'm talking to
RSTS it means it doesn't support X, so do Y instead". In the case of DAP this is
really blatantly wrong given the massive number of feature bits that spell out, in great
detail, what the other end is able to do.
This is actually a general protocol rule: never use implementation version info to make
protocol decisions; only ever use feature capability flags or the like. If those flags
are not adequate, get the protocol fixed to make them so.
paul