WORST keyboards ever

From: J. David Bryan <jdbryan_at_acm.org>
Date: Wed Sep 1 10:46:05 2004

On 30 Aug 2004 at 7:20, Jay West wrote:

> Actually, it probably would have worked, as a common (although not
> required) setup for HP terminals was that the serial interface in the
> host was set for external baud rate, and the terminal provided the
> clocking.

Indeed, it did work. The 264x terminals had a rotary switch that set the
baud rate clock supplied to the interface from 9600 down to 110. It had a
final position, "EXT," to set the baud rate from a clock received from the
interface (though I don't recall ever seeing a terminal in use with this
position set). Some interfaces, such as the 12880A CRT interface, set
their communication rate from the clock supplied by the terminal.

With such a setup, one could slow down a terminal listing by rotating the
switch to a lower baud rate. Because the switch was a break-before-make
type, one could even "tease" the switch into an intermediate position
(e.g., between 9600 and 4800 baud) to freeze the listing.

But I don't know why anyone would. Two of the primary operating systems of
the day, DOS and RTE, had terminal interruptibility (and, although I'm not
certain, I seem to recall that TSB supported XON/XOFF). Pressing any key
while terminal output was in progress would pause the output between lines
and issue a prompt for a system command. So pausing a long listing was a
matter of hitting SPACE to stop and RETURN to resume.

The only times I ever recall using the baud rate switch to slow down the
output was when running the hardware diagnostics. Even then, a few
characters were typically lost during the switching, showing up as filled
boxes (the DEL character graphic) on the screen.

                                      -- Dave
Received on Wed Sep 01 2004 - 10:46:05 BST