On 2/6/02 at 5:36 PM -0500, Cyrus Daboo wrote:
Yes indeed! I've already seen instances of a client that emits
'Thread-Topic' and 'Thread-Index' headers that I can't find
described anywhere. These are clearly meant to be an attempt to do
some form of smart message threading, which is fine except that it
doesn't use In-Reply-To or References, which is what rfc2822
suggests. So is this an attempt to deprecate the use of In-Reply-To
and References? This illustrates Keith's point: should someone be
allowed to register a header and state in the description that this
is meant to deprecate an IETF standard header?
Actually, I think this is a fine example. Let's assume for the sake
of argument that your evaluation is correct and this is an attempt to
do smart threading which could otherwise be done with
In-Reply-To/References. Whoever the bonehead was who came up with
Thread-Topic and Thread-Index, they are most certainly *not* going to
even attempt to write an Internet Draft documenting it: People of
this brain capacity can't imagine it's worth the time. And so off
this bonehead went and implemented it without comment. Guess what
follows: Other boneheads are going to go off and, having found no
documentation, figure out how to use it and are going to implement it
too. The net result of this little exercise is that over time we will
end up with a deployed stupid idea.
What's the alternative? Make a registry available for the bonehead to
register Thread-Topic and Thread-Index. Let it point to a web page of
his making that explains what a brilliant and intelligent person he
is for coming up with this wonderful mechanism: People of this brain
capacity think that reserving a name in a registry, thus
immortalizing themselves and publicly pointing out their brilliance
are fine things to do, especially if it takes no time at all. So the
fields get registered. Now, before this stupidity has a chance to
spread too far, those of us who are only mildly clueful see the new
registration. This gives us an opportunity to quickly try to convince
the bonehead that he is in fact a bonehead and get him to instead use
In-Reply-To/References and, if that fails, quickly come to a
consensus that this *is* a stupid idea and quickly get "This is
stupid; don't do it" put into the registry.
Giving folks a quick path to register field names and therefore a
quick way to look up field names gives the IETF community a chance to
communicate with people who would not otherwise hear from us before
they do damage. (Something like Dan's web page, though of high
utility, takes someone like Dan (who has a job other than
registrations) to go and post the new entries and is not easy to find
in the way that an IANA web page would be.) Giving folks a high bar
for registration does exactly the opposite of what you want: It makes
people ignore us, go off and do their own thing out of expedience,
and ends up doing just the kind of damage to the e-mail system you're
trying to prevent.
pr
--
Pete Resnick <mailto:presnick(_at_)qualcomm(_dot_)com>
QUALCOMM Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102