Re: Strict SMTP (RFC-821) compliance or not ?

[Felix Lee <flee@guardian.cse.psu.edu>]

> I believe the best policy is to be as lenient as possible in what you accept
> (although it is fair to warn the submitter of problems), but to be strictly
> compliant in what you generate.

This often-quoted guideline has a lurking contradiction, when you
start to consider more than two participants in a transaction.

example: X gives Y a badly formed message which is meant for Z.  Since
Y is lenient in accepting, Y accepts the message from X.  But since Y
is strict in generating, Y can't give the message to Z (unless Y
deletes or fixes the badly formed parts, which is not always safe to
do.)

Sorry to be unhelpful, but people repeat the above advice, apparently
believing that the right way to do things is immediately obvious.  But
I rarely have any idea what the immediately obvious behavior is
supposed to be:-)
--