At 19:02 -0400 02-10-03, Bruce Lilly wrote:
The suggestion was to relegate empty domain literals to an obs-
contruct, which means that parsing would not change, but it would
be come a MUST NOT generate construct. There need not be any
syntax checks for any parser w.r.t. this issue; any software
generating a domain literal would need to take some appropriate
steps.
Moving it to obs- syntax doesn't kill it, but it does declare it
as a wart.
One of the IETF golden rules is "Be conservative in what you
produce, be liberal in what you accept". According to this
rule, it is obvious that certain uses of empty strings are
things you should accept in input, but avoid producing.
I have a long time complained that IETF standards did not
tell implementors how to practically implement this golden
rule. What is the difference between the conservative
standards interpretation (what you should send) and the
liberal standards interpretation (what you should accept).
The "obs-" syntax of RFC 2822 is an attempt to codify this,
which I think is good. However, one can discuss whether
all "obs-" constructs have to be "MUST NOT". In some cases,
it may be reasonable to allow production of them. Whether
this applies to empty strings, I am not sure.
--- ---
In early mail systems, it was allowed to specify a
recipient address with only the local name. This would
then default to the sender's domain. Thus,
From: John(_at_)duckville(_dot_)org
To: Mary
would default to
From: John(_at_)duckville(_dot_)org
To: Mary(_at_)duckville(_dot_)org
This is not allowed any more, but some local UA-MTA
might still allow this technique.
--
Jacob Palme <jpalme(_at_)dsv(_dot_)su(_dot_)se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/jpalme/