[comp.protocols.ibm] BSMTP problem with UREP 3.2

LEONID@TAURUS.BITNET (04/20/88)

The problem I am going to describe below has occured since we installed
PSU UREP version 3.2 with it's BSMTP code. This has occured mostly with
old versions of Columbia (crosswell) MAILER running on VM, and lately
it occures with CUNYVM too.

It seems that the UREP code does not understand the full command set
of BSMTP and upon receiving something it cannot interpret it drops the
message and send a REPLY file back to the offending MAILER. I have not
succeeded to capture the original BSMTP command sequence that caused the
problem, but at the end of this message I supply the REPLY filw.

The current situation is that mail send by CUNYVM (and others)
to TAURUS.BITNET directly get dropped, and mail send to MATH.TAU.AC.IL
works fine since it passes through TAUNIVM's MAILER.

As it seems to me, the solution would be:
a) to fix the UREP code (!)
b) to remove extravagant commands and parameters from FAl and Columbia
   Mailers.

Of course a) is preferrable, but I can do it, the UREP code is totally
undocumented, not a single line of comment!

I would appreciate any help. Please reply directly to me:

leonid@taurus.bitnet
leonid%math.tau.ac.il@cunyvm.cuny.edu

----------------------
HELO TAURUS.BITNET
VERB ON
MAIL FROM:<>
RCPT TO:<mailer@cunyvm.bitnet>
DATA
220 taurus.bitnet Batch Simple Mail Transfer Ready
250 OK
250 OK
504 Command parameter not implemented
503 Bad sequence of commands
354 Start Mail input; end with <CRLF>.<CRLF>
503 Bad sequence of commands
221 taurus.bitnet closing transmission
.
QUIT
---------------------------

patterso@hardees.rutgers.edu (Ross Patterson) (04/25/88)

Leonid,

   The failing BSMTP command appears to be the "MAIL FROM:<>" command,
which is hardly gratuitous. It is persfectly reasonable SMTP as well
as BSMTP, and means "Under no circumstances are you to attempt to
reply to this mail. Should you find an error, simply drop the mail
without comment." It is also well known that some mailers (including,
I belive, the Crosswell) don't like this form. Someone from the
Weizmann Institute posted a fix for the Crosswell code to the XMAILER
list some time ago, perhaps UREP needs a similar fix?

Ross Patterson
Rutgers University

owens@psuvax1.psu.edu (Robert Michael Owens) (04/26/88)

In article <8804251118.AA06988@jade.berkeley.edu>, patterso@HARDEES.RUTGERS.EDU (Ross Patterson) writes:
+ Leonid,
+ 
+    The failing BSMTP command appears to be the "MAIL FROM:<>" command,
+ which is hardly gratuitous. It is persfectly reasonable SMTP as well
+ as BSMTP, and means "Under no circumstances are you to attempt to
+ reply to this mail. Should you find an error, simply drop the mail
+ without comment." It is also well known that some mailers (including,
+ I belive, the Crosswell) don't like this form. Someone from the
+ Weizmann Institute posted a fix for the Crosswell code to the XMAILER
+ list some time ago, perhaps UREP needs a similar fix?
+ 
+ Ross Patterson
+ Rutgers University

the bsmtp router which comes as part of urep supports "mail from:<>".

history (1983) - originally only "trusted mailers/routers" or "local
user agents" were suppose to be able to send mail to/through a BSMTP
mailer/router/gateway.  i.e., BSMTP was originally NOT to be a user
protocol. BSMTP routers were suppose to have been written for the
different systems on bitnet. it never happened. however many users
were willing to wrap their own BSTMP so they could use one of the
bitnet gateways (i.e., psuvax1). it was impractical (and unwise)
to register each such user. instead a router/gateway is suppose to
check BSMTP mail from non trusted users to make sure they aren't
lieing about who the mail is from. many of the gateways on bitnet
simply don't do this. the code supplied with urep does. trusted
mailers must be listed in a file "DOMAINS" which also specifies
the domain naming tree.

owens