Importing binary files without removable storage nor non-bundled software (was: TKermitFTP
6 Jan
2005
6 Jan
'05
11:43 a.m.
From: "Tom Jennings" <tomj(a)wps.com>
Hi
In the past when I wanted a non-block binary transfer,
I often had some value that I used as an escape character.
I'd choose a low frequency value to improve throughput.
It is simple, If you want to send a command value, you
send the escape character, first. If you just wanted the escape
character, you'd send it twice. For 8 bit data, you
had 255 commands. EOF could be one of those.
It makes a simple protocol that is easy to implement in
code.
Dwight
> "Binary reads from a device are not
> allowed" is the message at 8753 in DR-DOS 3.41.
On Wed, 5 Jan 2005, Vintage Computer Festival wrote:
The question I haven't seen answered yet is,
"Well, why the hell not???"
The real answer is: there is no out-of-band signalling, so one of
the 128 in-band symbols (SUB, 26 decimal) is chosen to mean "END
OF FILE". It's borrowed from CP/M-80; I don't know where DR got
it; it seems un-DEC-ish.
("Out of band" signalling is, for RS-232 et al, the hardware
handshake lines. No one likes them, everyone complains like
babies, so they have essentially been deprecated.)
6 Jan
6 Jan
1:48 p.m.
On Thu, 6 Jan 2005, Dwight K. Elvey wrote:
In the past when I wanted a non-block binary
transfer,
I often had some value that I used as an escape character.
I'd choose a low frequency value to improve throughput.
It is simple, If you want to send a command value, you
send the escape character, first. If you just wanted the escape
character, you'd send it twice. For 8 bit data, you
had 255 commands. EOF could be one of those.
It makes a simple protocol that is easy to implement in
code.
For MSDOS etc, you have to escape SUB, and NUL. And a bunch of
others. Plus, you might not have an 8-bit channel, only 7. So now
you ahve to escape all those.
A is sent as A
Control-A is sent as &A
A with bit7 set is sent as #A
Control-A with bit7 set is sent as #&A
Wait! That's Kermit!
Machines need to pause for various tasks, like writing to slow
disks. So now you need a way to tell the sender to stop.
Ad nauseum. Hence, communications protocols. The problem is
anything BUT trivial.
But I do agree with you, I'm all in favor if one-off problem
solutions, with the caveat that many tasks expand with time.
To paraphrase the jargon file:
"That's why I love standards, there's so many to choose from!"
2:47 p.m.
In the past when I wanted a non-block binary transfer,
[...]
It makes a simple protocol that is easy to implement
in code.
It's _almost_ identical to SLIP framing (though SLIP has marginally
different goals, because its EOF analog occurs relatively frequently
(depending on the data stream, anywhere from every few dozen bytes to
every thousand or so). A couple of years ago, I even took advantage of
some of those 256 sequences that the escape-character technique gives
you to extend SLIP to handle IPv6. (The SLIP "nonstandard" contains
language implying that this cannot be done. Happily, that implication
is wrong - whoever wrote it apparently didn't notice that SLIP uses
only two of the 256 possible escape sequences and that 254 of them thus
remain available for other uses.)
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse(a)rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
3:20 p.m.
On Thu, 6 Jan 2005, Dwight K. Elvey wrote:
In the past when I wanted a non-block binary
transfer,
I often had some value that I used as an escape character.
I'd choose a low frequency value to improve throughput.
It is simple, If you want to send a command value, you
send the escape character, first. If you just wanted the escape
character, you'd send it twice. For 8 bit data, you
had 255 commands. EOF could be one of those.
It makes a simple protocol that is easy to implement in
code.
A perfectly good and simple protocol for a simple serial interface (i.e.
2 or 3 wires max). But I submit that the BREAK signal could have been
used, which is inline and is part of the RS232 spec for this very reason.
Anyway, dead horse is now a pile of fur. The ants are already back in
their hole and the flies have long since flown away.
--
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
[ Old computing resources for business || Buy/Sell/Trade Vintage Computers ]
[ and academia at www.VintageTech.com || at http://marketplace.vintage.org ]
11:26 p.m.
Vintage Computer Festival wrote:
A perfectly good and simple protocol for a simple
serial interface (i.e.
2 or 3 wires max). But I submit that the BREAK signal could have been
used, which is inline and is part of the RS232 spec for this very reason.
Or you could use bit-banging with the RTS and CTS lines, to send actual
out-of-band data and commands outside the "normal" serial link.
Ok, ok, it was just a thought ;-)
Gordon.
7208
days inactive
7209
days old
4 comments
5 participants
participants (5)
-
dwight.elvey＠amd.com -
gordon＠gjcp.net -
mouse＠Rodents.Montreal.QC.CA -
tomj＠wps.com -
vcf＠siconic.com