[Simh] EXT :Re: PDP-10 simulation: DEUNA support help needed

[Rich Alderson]

> Date: Wed, 29 Apr 2015 22:10:10 -0400 (EDT)
> From: Cory Smelosky <b4 at gewt.net>

>> You would still have to assemble a file system by hand.  WAITS originated
>> before DEC's Level D disk code, and the systems programmers defined their own
>> file system.  PPNs are SIXBIT, so [1,2] is octal 21,,22 in the MFD entry (yes,
>> they're right justified in the halfwords).  The last version of the file
>> system, which lived on 3 RP07 disks, allocated things in blocks of 9 sectors,
>> giving 1K of data + 200 words of "retrieval data" per block (repeated through
>> every block of a file).

> Interesting!

> Similar to TOPS-10's FS with multi-unit structures or with some sort of 
> striping/parity?

Multi-unit structures.  Striping came along with redundant arrays of
inexpensive disks.  At $50,000 a drive, RP07s were not candidates. ;-)

>> The issue is that only 3 systems ran WAITS: SU-AI at SAIL, a KL-10 system
>> attached to the S-1 project at Lawrence Livermore Labs, and a Foonly F2 at
>> CCRMA (the Center for Computer Research in Music and Acoustics, at Stanford).
>> Because of the way the system grew, there was never a need for a utility that
>> would build a file system from scratch on an initialized disk (like CHECKD in
>> TOPS-20 or the ONCE code in Tops-10, or even NSALV in ITS), so no one ever
>> wrote one.  The closest thing to it was a program written to move from a
>> damaged file system to a new set of RP07s, which relied on working copies of
>> several programs on the undamaged disks.

> I take it from your understanding of the FS you have a utility to do just 
> that written? ;)

>> Getting a file system working was about 6 months of last year.

Long talks with two old friends, both SAIL alumni who worked on the OS when it
was a research specialty all its own, were the first part--and sometimes one or
the other would not remember changes made for later devices (like the RP07s).
Then I discovered the program RSKINI in browsing SAILDART.org, and things began
to gel; I eventually converted it from FAIL to Macro-10, instrumented the hell
out of it[1], and took the logged output as the source for the next part of the
process.

That consisted of setting up FILDDT batch jobs on Tops-10 (patched to not treat
foreign drives as offline) where the output from RSKINI was scribbled directly
onto the disks.[2]  Once that was done, WAITS could boot past the point where
it wanted to turn on swapping.

>> Me, too.  TENEX requires either a KA-10 based system with a BBN pager attached,
>> or a KI-10 based system (and DEC would sell you a license for their modified
>> version).  If I ever have time, I might try getting TENEX to run on our 1070.

> Does that mean you have bootable TENEX installation media, or would you 
> need to construct manually and cross-assemble before rebuilding itself?

I don't think anyone has bootable TENEX media.  It would be a labor of love to
try to cross-compile TENEX under TOPS-20 on a Toad-2, then build a boot program
for the KI that could read it in from disk.  At least the file system layout is
similar to TOPS-20.

We've probably bored the broader audience enough.  Talk to you soon in private
e-mail.

                                                                Rich