Replacing 822 and in the course silently dropping X-* would leave a lot
of software and systems stand outside without the protective umbrella
they had up to now; they survive but if it rains, they get wet.
Some grepping over a couple of inboxes showed that in 650 messages there
were 1232 X-* headers. Many anti-spam and anti-virus systems use such
headers, as does some mailing list software etc, even the messages from
this ietf-822 list do have a X-Authentication-Warning: header.
RFC-1918 is a godsend for IP addressing. X-* headers could be handled in
just a paragraph or two and can be a godsend in many situations as well.
Users and systems which need to use some proprietary header and can do
with an X-* won't need to bloat the standards with their specialized
extensions.
I think there definitely should be either a note in 3822, and a separate
specification, or (preferrably and simpler) a short definition directly
in 3822.
On Fri, Jan 02, 2004 at 11:55:43AM -0500, Keith Moore wrote:
We went through this before. At the end, noone raised objections against
the insertion of a sentence such as "X-* header fields are not covered
in this document". Please, can we just insert that in the list of
changes and be done?
Quarrelling about whether X-* is useful, a mistake, neither or both is
not going to help 3822, so let's not do it.
well, I do sympathize with the idea that being silent on the subject is
confusing. but I don't want that discussion to be critical path for 3822.