From: Tomi Manninen (cyysnu@tds.net)
Date: Sun Jul 15 2001 - 03:28:10 EEST
On Mon, 9 Jul 2001, Roger Harris wrote:
> Thanks for the reply, John. I think what you describe is a different
> problem, which I have also noticed.
>
> The problem you describe is a quirk of BPQ. It won't accept commands unless
> they are followed by a carriage return and (this is the nasty bit) the
> carriage return must arrive in the same packet as the command itself.
>
> This is OK if you come in from an AX25 node or a TNC, since the packet is
> not normally sent until the CR is received (or until the maximum packet
> length is received). However, if you are telnetting to a TCPIP system first
> (e.g. Windows or Linux), that system will most likely split your message up
> into AX25 packets, each containing a single character, especially if you
> are typing slowly. So the BPQ box will receive the first character of the
> command, without the CR, and immediately complain.
This is true. However the problem actually is deeper than you would think.
AX.25 is a "sequential packet" protocol. The packet boundaries are
actually important and meaningful in the transport of data. However most
use of AX.25 ignores the sequential packet nature of the protocol and uses
other means for delimiting the information. You use new-lines and ">"
chars etc. to denote that a "frame" of information has arrived.
LinuxNode doesn't respect the packet boundaries either. I wrote libax25io
(the I/O lib underneath node) to be a sort of a SEQPACKET to STREAM
converter. I have also recently added a feature to further "rape" the
packet boundaries in it, the primary reason being the boost it gives to
the stream compression of connections available in node (it defaults to
being off of course).
A whole another issue regarding the original topic is that the problem you
describe should only happen with braindead telnet clients like the windows
one. Node does try to negotiate with the client what is called LINEMODE
where data is only sent after a carriage return. If anyone is seeing a
char-by-char behaviour with any halfway decent telnet client then let me
know. That is not supposed to happen.
-- Tomi Manninen Internet: twhmishv@switch.com OH2BNS AX.25: terhi.victor@logonet.com KP20ME04 Amprnet: hhzdpk@tusur.ru- To unsubscribe from this list: send the line "unsubscribe linux-hams" in the body of a message to mlrougnw@flmultimedia.com
This archive was generated by hypermail 2b30 : Sun Jul 15 2001 - 03:51:06 EEST