From: Tomi Manninen (cxbz@them.net)
Date: Sun Jul 15 2001 - 03:55:42 EEST
On Mon, 9 Jul 2001, Roger Harris wrote:
> 1) FBB - telnet - FBB This works OK. Not too surprising, since the two
> FBB systems are both handling Telnet in the same way.
>
> 2) FBB - ax25 - linuxnode - telnet - FBB. Here, the compressed forwarding
> falls over. The Telnet session is being originated in the linuxnode and
> although the escape character has been disabled, the node is adding
> linefeed and additional 0xFF characters which are not expected by the
> receiving FBB.
>
> 3) FBB - ax25 - linuxnode - telnet - linuxnode - ax25 - FBB. This also
> works OK. The first linuxnode is adding linefeeds and 0xFF characters, but
> the second node is expecting this, and strips them off.
Ah. Thanks. This kind of explanation of the problem definitely helps. I
don't need to guess what was done when the bug happened...
One question though. Have you tried "FBB - telnet - linuxnode - telnet -
FBB" ? The thing I mentioned in my earlier email (that would probably
need fixing in node) was related to this situation.
> So in summary, it is an incompatibility between the way FBB and Linuxnode
> handle telnet sessions.
Most probably, yes.
For background information the 0xFF character is concidered IAC (Interpret
As Command) in the telnet protocol. The characters after it are
interpreted as telnet protocol stuff up until a certain char sequence
depending on the command. To send a 0xFF over a telnet connection it need
to be doubled (an "IAC IAC" sequence is interpreted as a single 0xFF).
However I think FBB handles this as well. The problem with compressed
forwarding might be the end-of-line handling included in node. With most
TCP based protocols on the internet the EOL is the CR-LF sequence. Also in
most situations using the TELNET protocol (which is usually run over TCP)
the EOL is either CR-LF or CR-NUL, a two-byte sequence.
I need to think about the EOL handling a bit more and see if I can come up
with a solution that would satisfy most people. Jean-Paul if you have any
thoughts on this, please let me know.
> However, it does appear that this is a very widespread problem and has been
> confusing a lot of people for some time. I have noticed some IP systems
> using plain text forwarding, since they could not get compressed forwarding
> to work. The problem is not limited to Linux. People have reported similar
> problems with JNOS, TNOS, etc. I suspect there are similar incompatibilities.
Yes, long time ago I fixed a number of problems related to this in the
JNOS mbox feature. That was stuff ranging from the mbox escape char to the
handling of internal data structures in certain complex cases of input
data... :-) Quite a lot of time was spent of all that... :-)
-- Tomi Manninen Internet: lnxsapba@wjla.com OH2BNS AX.25: terhi.victor@logonet.com KP20ME04 Amprnet: vxkk.jmfbrcen@trs.su- To unsubscribe from this list: send the line "unsubscribe linux-hams" in the body of a message to lba@roskapostia.tunk.net
This archive was generated by hypermail 2b30 : Sun Jul 15 2001 - 03:56:20 EEST