Re: #1417 USEPRO 3.4: Injecting-agent modification of message ID

["Charles Lindsey" <chl@xxxxxxxxxxxxxxxx>]

In <JC2Kzt.J1p@xxxxxxxxxxxxxxxx> "Charles Lindsey" <chl@xxxxxxxxxxxxxxxx> writes:

>In <87r6ttpoz0.fsf@xxxxxxxxxxxxxxxxxxxxx> Russ Allbery <rra@xxxxxxxxxxxx> writes:

>>Charles Lindsey <chl@xxxxxxxxxxxxxxxx> writes:

>>> No, I think SHOULD is sufficient. As usual, it means "there needs to be
>>> an exceptional reason to break it". I can't think of such a reason
>>> offhand, but that is not to say it cannot arise. OTOH, there appears to
>>> be no interoperability problem, unless you want to count the existence
>>> of two articles with the same text as interoperability.

>>There is a significant interoperability problem in that we support
>>simultaneous injection with multiple injecting agents and rewriting the
>>message ID breaks that.  This is, in practice, not an uncommon thing to
>>want to do.

>>Message-ID here is special because of that.

>A valid point. I still mildly prefer remaining with SHOULD, but accept
>that I seem to be outvoted.

I have been thinking further about this, and the consequences if various
headers DO get changed by injecting agents and the article is then
injected in two places.

Message-ID:

If that case you have, technically speaking, two distinct articles which
just happen to be otherwise identical, and they will both propagate around
Usenet quite happily causing no technical harm whatsoever. Of course, it
will annoy people (and indeed it often happens currently, usually because
someone thought his article had failed to get out, and pressed 'send'
again - usually resulting in mild flames in his direction).

Date:, Path:, Sender:, User-Agent (and anything else an injecting agent
might be tempted to change)

In that case you have two different versions off what is supposed to be
the same article (i.e. they have the same messagg-id). Some sites will see
one and some the other. That is not supposed to happen in a well-regulated
network. OTOH, no huge harm arises, so "SHOULD NOT alter" (as in the
present text) may be sufficient. It just strikes me as slightly odd that
we are proposing a "MUST NOT alter" for the case that causes no technical
harm (but much annoyance, because people will notice, which they would not
in the other cases).

Xref:, newly added bits of Path:, Injection-Info (and maybe Injection-Date:
according to the outcome of #1416)

And these, of course, are actually intended to be different.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl@xxxxxxxxxxxxxxxx      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5