🚀 go-pugleaf

[ I don't think I'm saying anything new in this post.  Every time this
header gets debated people say the same things.  I've listened to the
discussions before, and now I'm going to try to post what I feel is a
minimal consensus compromise.  Henry can take it or leave it. ]

I would like to propose the following compromise:

In the son of RFC 1036 (is that the right number?  I'm a bit sleepy
tonight), I with that only the following two forms were permitted:

Distribution: real.news.group.name

and

Distribution: word

"real.news.group.name":  Passes the message if the neighbor
would get this newsgroup.  (this won't slow things down since
all modern software uses a fast function to answer this question).
[ i.e. For posting to news.answers,alt.sex but restricting it
to only where alt.sex is found.  Or, posting to comp.misc.unix
but only where comp.lang.perl is found (hey, what a great way
to hide from the "DOS-perl" weenies!) ]

"word": Passes the message if the distribution is listed
in the distribution field of sys or newsfeeds or whatever.
[ i.e. the flat namespace that we all love to hate. ]

These two things are compatible with most news software.
(or could be added to INN very quickly.)

What would be "quietly" removed from the RFC (or the unwritten list of
bizarre RFC interpretations that news systems have propagated.)

- attempts to use a hierarchy (foo.bar, when there is a foo.bar.bat but
  no foo.bar) as a distribution.
- attempts to use multiple words in the Distribution: header or "!" and
  other pseudo-operations.
- attempts to use the "big seven" (comp,misc,...) as distributions.

Why reduce the functionality?  It reduces "Distribution:" to the
most basic functionality.  Using "Distribution:" for anything
fancy just doesn't work.  This is always the conclusion when
this header is discussed in any technical forum.

Why not fix the functionality?  Because nobody has been able
to "push the envelope" successfully beyond what was listed
above.  Any hopeful ideas resulted in total incompatibility.

AN INCOMPATIBILITY??? WE CAN'T HAVE THAT HERE!! THIS IS USENET!!!
Yes, I agree (ha ha, only serious).  If people want to "push the
Distribution: envelope" they'll have to invent a new header.  Doing so
would not be too difficult.  The difficulty would be to define what
problem needs to be solved.  In the past couple of years I've posted a
number of "solutions looking for problems" that would be good
candidates for a new header.  However, I don't feel that anything new
should be invented in the son-of-1036 RFC, as I want to see it get
finished some time this year.  After it is done, and people start
talking about completely new features for a son-of-son-of-1036, then
I'll repeat the suggestions that I've made/heard in the past; some will
get tested; and the ones that work will be added to
son-of-son-of-1036.

What do y'all think?

Tom
-- 
Tom Limoncelli -- tal@warren.mentorg.com (work) -- tal@plts.uucp (play)

Unix is a trademark of, WHATT????