Re: Please publish attached draft-ietf-secsh-break-01

[Ivan Passos <ivan.passos%cyclades.com@localhost>]

Bill Sommerfeld wrote:
> 
> > Kermit software has supported since the early 80s both a Short Break and
> > Long Break.  A break less than 300ms is short; a break greater than
> > 700ms is long.  These due have meanings for some modems and some
> > terminal servers.
> 
> I assume there's also a minimal length for "short break".

A serial break is defined as the assertion of the TxD signal to 
'0' (Space) for a period longer than a character time. 

So, for 9600 bps, the minimal break length would be:

	1 / 9600 x (8 data + 1 start + 1 stop) = 1ms

For 115200bps, the minimal break lentgh would then be 87us (!!!).

So, to guarantee the serial break works in _any_ baud rate, 
communication programs usually define the break length as larger 
than any baud rate possibly used. Since in the old, old, old days 
people actually used 50bps as a valid baud rate, then the minimal 
"universal" break would have to be larger than 200ms.

That's where the 300ms definition in Kermit came from.

And yes, some UARTs are capable of detecting the break length 
(not just the break condition) and differentiate between 
short breaks and long breaks. There is no standard as to what 
exact length makes a break short or long.

Sorry if this has been just boring, useless information, but 
I thought I should mention it. :o)

Later,
-- 
Ivan Passos			Phone: 510-771-6202
Product Marketing Manager	Cell:  510-468-3515
Cyclades Corporation		Fax:   510-771-6200
ivan.passos%cyclades.com@localhost
http://www.cyclades.com		"Everywhere with Linux"