From: Brian Palmer (bpalmer@leland.stanford.edu)
Date: Wed Mar 06 2002 - 07:29:57 CST
chl@clw.cs.man.ac.uk (Charles Lindsey) writes:
> OK, it now says
>
> 10. The injecting agent MUST NOT alter the body of the article in
> any way. It MAY add other headers not already provided by the
> poster, but SHOULD NOT alter, delete, or reorder any existing
> header, with the specific exception of "tracing" headers such as
> Injector-Info and Complaints-To, which are to be removed as
> already mentioned.
I like it.
[snip my wording which explicitly deprecated the practice]
> >I could live with that. However, *if* the sender's mailbox is being added
> >in accordance with local policy, surely Sender is a better place for it ?
> >Try:
>
> > in particular, adding a Sender header that exposes a sender's
> > mailbox has privacy implications, and it may be more desirable
> > to add tracing information via the Injector-Info header (see
> > 6.19.1 for usage notes on Injector-Info).
>
> I think Brian Palmer's "deprecated" was a bit too strong, but I think your
> "more desirable" is too weak.
I am not sure that deprecation is quite the best way to handle it, but
it seems to have many of the connotations you're looking for (works
now, new applications should not use it and as software is updated
that approach should be abandoned). We deprecate other things in the
current draft, so it's not a novel concept, either.
> Essentially, we have some existing practice which we do not entirely like,
> so we would like to apply gentle pressure to have that practice
> changed.
>
> What I now have is:
>
> NOTE: The addition of non-mandatory headers by the injecting
> agent may alter the posting agent's preferred presentation of
> information. Ideally, the injecting agent will comply with the
> poster's preferences; in particular, adding a Sender-header that
> exposes a sender's mailbox has privacy implications, and it is
> preferable to add any such tracing information via the Injector-Info
> header as set out in 6.19.1.
I've had the chance to think a bit more about it, and this is still
dancing around the point, by trying to be a bit more general and a bit
vaguer than is useful (and yes, I do recognize that I'm arguing in
part with my earlier suggestion). John Stanley raised the good point
that the purpose of the standard is to express the format, not just
"encourage people." If it is in fact *always* preferable, we should
just make it a 'SHOULD NOT add Sender'.
Specific phrases I now dislike are the 'ideally...',and the 'privacy
implications' phrase, as it's very non-specific (and thus of little
help to an implementor).
...may alter the posting agent's preferred presentation of
information. Of especial note, adding a Sender-header exposes a
mailbox to the article's recipients; information regarding the
sender may better be placed by the injecting agent in the
Injector-Info header as set out in 6.19.1.
The problem with the 'ideally...' section is that it is mismatched
with 'MAY add other headers' (and a few other MAYs in the text). Of
course, the NOTE is non-normative, but I think it's confusion for no
real purpose.
-- Brian Palmer "Whoever fights monsters should see to it that in the process he does not become a monster. And when you look long into an abyss, the abyss also looks into you" - Nietzsche