[Simh] PDP paper tape reader end of file conditions

[Bob Supnik]

DEC didn't make much investment in error detection for its paper tape 
equipment. In the 18b family, the PDP-9 was the first 18-bit system to 
feature a "reader empty" flag. The PDP-8 never had one. (The PDP-11 had 
one from the get go.)

That begs the question of what these early PDP's did when they ran out 
of tape. The reader logic was very simple; when the reader saw a 0->1 
(dark to light) transition on the feed hole, it strobed a character. So 
there might be a last garbage character when the tape ran out, but after 
that, the feed hole always saw light, and there would be no further 
transitions.

Some drivers used timing loops or the clock to "time out" the reader. 
The readers ran at 300cps, so if say 10-20ms went by without a 
character, end of tape was a good guess. Another strategy was to 
"guarantee" or require some sort of end of input marker on the tape. For 
example, hardware read-in mode terminated when it detected a punch in 
channel 7.

The PDP-1 presents different problems because it has IO modes that are 
absent from later PDPs - synchronous and asynchronous wait. These 
allowed programmers to issue a read and then stall (immediately or 
later) until a character was ready. The problem, of course, is what 
would happen if the reader ran out of tape in these modes. My best guess 
is that the PDP-1 would stall until reset.

This came up during debug of Expensive Typewriter, the PDP-1's editor. 
If an input tape isn't properly formatted and runs off the end, ET hangs 
up in an indefinite stall. When ET writes its output to tape (via p), 
users must give another command at the end (s) to write the appropriate 
end of tape marker.

/Bob