Hi Rob,

Rob Swindell wrote to Michiel Broek:
RS>   Re: Telnet and zmodem
RS>   By: Michiel Broek to Rob Swindell on Mon Mar 06 2006 02:54 pm
RS> 
RS>  > Hi Rob,
RS>  > 
RS>  > it seems that the problems between syncterm and mbse are in the 
RS> underlying
RS>  > network part. Mbse uses the telnet links as 8 bit link, syncterm 
RS> escapes
RS>  > telnet control commands. I experimentyed with that too, but they two
RS>  > programs still refuse to work together.
RS> 
RS> When downloading via Zmodem, the upstream packets are all in ASCII, so I 
RS> don't
RS> see how "escaping telnet control commands" in SyncTerm
would matter. Of 
RS> course
RS> SyncTerm expects the Telnet IAC character (ASCII 255) to be
"escaped", 
RS> that's
RS> how the Telnet protocol works. Is it this escaping of the IAC that you're
RS> saying MBSE does *not* do?

The upstream seems not the problem, downstream. And yes I don't escape IAC
because on IP connections mbsebbs is run from inetd -> in.telnetd
So mbsebbs doesn't answer the incoming connections itself. In my opinion
in.telnetd handles the IAC escaping. The only thing mbsebbs does is put the
tty into raw mode using ioctl calls. And up/downloading data containing
character 255 is possible.

RS> 
RS>  > I did some more testing and made the following matrix:
RS>  > 
RS>  > 
RS>  >                 Mbsebbs       Synchronet    Concord      Mystic
RS>  >                 ------------- ------------- ------------ -------------
RS>  > Dynacom-8 win   Ok            Ok            Error        Aborted
RS>  > mtelnet win     Ok            Ok            Ok           Ok
RS>  > Minicom Lnx     Ok            Ok            Ok           Ok (crc 
RS> errors)
RS>  > Syncterm lnx    Error         Ok            Ok           Aborted
RS>  > TeraTerm win    Ok            Error         Ok           Error
RS>  > 
RS>  > All were tested with a large download using normal zmodem.
RS> 
RS> Of course Synchronet doesn't have any built-in x/y/zmodem transfer 
RS> protocols so
RS> when you list "Synchronet" above, that's too vague to be
useful. Depending 

RS> on
RS> the BBS which you connected the sending Zmodem protocol driver could 
RS> easily be
RS> FDSZ, CEXYZ, [l]sz, or SEXYZ.

I tried vert.synchro.net for this and selected Zmodem. Was that lsz?

RS>  > To me it seems there are probably tiny differences which break some
RS>  > combinations. We may better not discuss Mystic, it's possible that 
RS> there is
RS>  > something wrong on the bbs side of that system, only mtelnet did ok 
RS> without
RS>  > errors, other programs finished ok with lots of crc errors or aborted 
RS> on too
RS>  > many crc errors.
RS>  > 
RS>  > Do you have any ideas what the real problem may be between these
RS>  > incompatibilities?
RS> 
RS> Not without more details. :-(

That's the problem, there still are no real details. But the new logging in
syncterm made a lot more clear already. I also have a cvs checkout here now
so I can make changes for myself if I need that.

My next plan is to look at the in.telnetd sources to see if something
therein does something I don't expect.



    Greetings, Michiel Broek

Email:   mbse{at}mbse.dds.nl
Fidonet: Michiel Broek at 2:280/2802

... Going the speed of light is bad for your age.

--- MBSE BBS v0.83.15 (GNU/Linux-i386)