[svlug] Email length

Derek J. Balling dredd at megacity.org
Mon May 15 12:59:02 PDT 2000

At a fundamental level, I do not agree with the MTA caring about the 
content of the DATA segment at any time. However, I can also acknowledge 
that this is an area that no good RFC has been written to cover yet, so I 
can accept that there may need to be some flexibility on this front in the 

BUT -- There is a difference between ALTERING the content, and SCANNING the 

If the MTA is going to accept a message, it should accept it as-is with no 
modifications (except to the headers as specified in the appropriate RFC). 
If the MTA decides - for whatever reason it chooses - to deny the message, 
then it may do so, but MUST do so in the appropriate manner: sending a 
500-series response code to the sender indicating permanent message failure 
and the reason for such failure. The sender MUST be aware of the status of 
the delivery success/failure. Dropping the message on the floor is not an 
acceptable solution.

However, there are a few caveats:
1.) There are some (poorly made) MTA's which do not properly handle 
receiving a 500-series failure AFTER they have delivered the DATA segment. 
Depending on the MTA in question, some will not see it as an error, but see 
it as success and not generate a DSN for the sender. They may also 
continuously retry to send the message. (Not seeing it as "Success" but 
missing that it is a "failure")

2.) Most schemes designed to find attachments in messages are not too 
well-thought-out and will false-positive more than is proper.


At 12:48 PM 5/15/00 -0700, Ray Olszewski wrote:
>I recall some discussion here, a week or so ago, about using filtering
>rules, at the MTA level, to block the local delivery of "active content"
>components of messages ... the context was as a way of blocking the spread
>of things like the "ILOVEYOU" worm. Do you feel that such actions are
>equally inappropriate?
>At 12:33 PM 5/15/00 -0700, Derek J. Balling wrote:
> >At 09:44 AM 5/15/00 -0700, Emmanuel-Robert Mayssat wrote:
> >>A better way to enforce 76 char/line would be to configure the MTA
> >>( mail server ) to do the job.
> >
> >No. An MTA must _NEVER_ dick with the contents of the message. The MTA does
> >not know what standard is necessary for the content contained within, nor
> >SHOULD it care. It is a TRANSPORT agent, just as the acronym states. It has
> >nothing to do with CONTENT whatsoever.
>------------------------------------"Never tell me the odds!"---
>Ray Olszewski                                        -- Han Solo
>Palo Alto, CA                                    ray at comarre.com
>svlug mailing list
>svlug at lists.svlug.org

More information about the svlug mailing list