🚀 go-pugleaf

I missed some of this thread, but I remember that most operating
systems had a padding factor you set for each terminal, the number of
NULs to follow a CR.  We got a lot of throughput improvement by
simulating the carriage position and sending the minimum number of
NULs required to return it from where it was. Since it gave the
illusion of faster response, the powers-that-be let us expend compute
cycles on it (we finally determined that the cost of not processing
the interrupt-completes on the unnecessary NULs paid for the
additional cycles required to maintain the simulated carriage
position).

We once had a plotting program that drew graphs using characters on
the terminal.  I wrote an optimizer for it that took into account the
carriage-return time, backspace time, and tab time.  It put out a
little calibration sequence (if you needed it) that put a bunch of
vertial bars out using spaces between, then would move the carriage
over, and send a HUGE number of NUL bytes, telling the user to push
the "set tab" button (tabs were implemented by physical hardware on
the 2741), then draw the same again with tabs and vertical bars.  It
would then ask for confirmation that the tabs were all set.  Once it
had this information, it would try to optimize each line relative to
the last one by using the fastest drawing sequence possible. I usually
got a factor of 3 or 4 output speedup; drawing the axis on the left
was always the bottleneck.  Then I discovered the "inverse line feed"
character code and started rewriting it to take advantage of this, but
never finished that project (the optimizations got a lot hairier).
				joe

On 10 Feb 1998 19:53:51 GMT, an...@rmi.net (Andy Rabagliati) wrote:

>According to Norman Brake  <br...@austx.tandem.com>:
>> 
>> Hmm...  As I recall, there was a dash-pot on the left end of the
>> carriage of ASR-33's.  If its vent wasn't properly adjusted, either
>> the carriage wouldn't get back to the left end in time (and more
>> NUL's would be required) or it would bang the left end too hard.
>
>And heaven forbid that your paper tape somehow got into the
>dashpot. It could jam the plunger so tightly you needed to
>take the thing apart.
>
>Cheers,       Andy!

Joseph M. Newcomer
newc...@flounder.com
http://www3.pgh.net/~newcomer