1 Jul
2022
1 Jul
'22
9:22 p.m.
Early use of ASCII control characters was ALL over the map.
I remember typing in a BASIC program to approximate PI during a
college tour in the late 70's, and whatever system they had (maybe
Univac?? there was an unplugged RCA Spectra in the hallway) required
typing Control C (ASCII ETX) instead of RETurn on the teletype.
*THAT* was a mind-f*ck!
Clem Cole wrote:
Steve - The social disease spread of DOS-11, RT-11,
CP/M, and MS/PS-DOS
used ^Z as an EOF character in their text file format. The key is that
they stored a block count, not a byte count in the META. Thus the last
byte needs a marker to tell the OS to stop reading. [Early DEC OS's may
have done that also, but I never looked at their FS formats].
I TEND to think of the DEC ASCII traditions as coming out of the PDP-6
Monitor (later TOPS-10), which probably dates to c. 1964, and modern
Unix/Linux systems are almost fully infected with DEC conventions
(DELETE, CTRL/U, CTRL/C vs @/#/DELETE for example), except for ^M and ^Z.
The earliest version I have on hand is version 3.19, which includes
PDP-10 (KA10) support (c.1968). The scheduler (CLKCSS) has Tom
Hastings name on it, dated 6 SEP 67. Tom wrote the MIT CTSS
(timesharing on IBM 709/7090) scheduler, and the DEC T/S system
mirrors CTSS in many ways (project/programmer numbers). CTSS used
ASCII, and it would be interesting to know what conventions were used
there!! On topic: Ken might know. @/# seem to have been used on
Multics.
The full duplex terminal code (SCNSRF) with Robert Clemmens name
on it, dated "27 NOV 68" has the following conventions:
Use of "^Q" as a convention for control characters.
CR ^M return to column zero (used alone in files for overstrike)
LF ^J line feed: line terminator
HT ^I horizontal tab
VT ^K vertical tab (n line feeds on a TTY)
FF ^L form feed
XON ^Q start TTY paper tape reader
XOFF ^S sent by TTY PTR on end of tape?
^U clear input line
^Z EOF
0177 DEL delete a character
033,0175,0176 "ALTMODE" -- used in DDT (& TECO), echoes as "$"
I don't have a set of Commonly Used System Programs (CUSPs) from that
era, so I can't see how they handle ^Z if they see it in a file.
I don't remember seeing ^Z in disk files on TOPS-10.
A quick look at the 3.19 code (I rescued it from six DECtapes while at
DEC; three each for full & half duplex versions of the code!) and
didn't spot the on disk/DECtape metadata formats, but I suspect they
were just block counts. There ARE files I've pulled off of tape with
embedded NULs, so I suspect code was written to ignore NULs whereever
they appeared. "Dump Mode" I/O, where you passed IOWDs -COUNT and
WORD_POINTER in halfwords, rather than transferring character at a
time into a buffer, was a way to speed up disk I/O, which would easily
result in multiple embedded NULs. My recall is that we thought of
disk files as made up of blocks, and not characters.
MIT AI's Incompatible Timesharing System (ITS), a dig at CTSS (and DEC
as well?) used ^C instead of ^Z for EOF, appearing (in triplicate?)
inside of (older) files. And ^Z stop a (sub)job and drop into DDT (the
debugger *AND* command interpreter), which could be examined,
continued, or killed (ALT CTRL/X). I've always assumed BSD used ^Z as
the suspend character for this reason.
TOPS-20 (and perhaps later day TOPS-10) had file metadata
indicating file byte counts, and byte sizes (1-36 bits), but
TOPS-20 ALSO only allocated file storage on a page basis (512
36-bit words instead of the TOPS-10 256W block size).
The PDP-7 unix inode stores a word count, rather than a byte count,
and ISTR programs ignore NULs for the same reason.
Unix, of course, never made any distinction to the
core OS WRT to 'type'
[other than Regular/Directory/Special] and Ken stored a character count.
So there was no need to signal EOF with a markered stored on disk..
A pipe or the shell on the other hand does have a need to signal the end of
a transaction, and 'End of Transmission,' as Nelson points out, is the
ASCII character reserved for the same.
I don't know the rationale for DEC's use of ^Z for EOF, but as I
pointed out above, use of ASCII control characters wasn't widely
standardized in the 70's (is it today?)
To pull the thread back to the list topic
1. Remember which of @ and # deletes a single character.
2. Fix a filesystem without fsck.
3a. Buy a controller card to be able to use a disk
3b. Write a driver for a third party controller.
4. Edit a driver and recompile the kernel to change partition sizes.
5. Use dd to convert a mainframe tape to variable line length ASCII.
6. Wonder whether something with a serial port is a DTE or a DCE.
7. Figure out the serial port settings for a terminal/modem:
data bits, parity, stop bits.
8. Adjust jumpers on a peripheral
9. Insert or remove wires for IRQ bus continuity
Non-UNIX related:
Punch job control cards.....
Burst fanfold paper by flicking their finger, to get your printout.
Spread out fanfold paper on the floor to get a "big view"
.......