From: Charles Lindsey (chl@clerew.man.ac.uk)
Date: Tue Mar 16 2004 - 06:13:27 CST
In <405661AC.4000202@erols.com> Bruce Lilly <blilly@erols.com> writes:
>Charles Lindsey wrote:
>>
>> The In-Reply-To header is not, and never has been, a part of Netnews.
>Both RFC 850 and RFC 1036 use RFC 822 as a basis, and In-Reply-To
>is defined in RFC 822; there is nothing in RFCs 850/1036 that
>precludes use of In-Reply-To. In-Reply-To may be found in Usenet
>articles (if you look, you will find it).
It is generally the case (and our draft confirms this) that any header
defined for Email MAY be included in a Netnews article, and if any agent
can make use of the information in such a header, it is welcome to try.
But there is no requirement to generate the In-Reply-To header, nor has it
ever been suggested that there should be.
>Except for the trivial case of a followup to multiple messages,
>where each of those earlier messages has no predecessors, the only
>efficient way to provide reliable indication that the message is in
>fact a followup to multiple predecessors is via In-Reply-To or
>something equivalent to it.
>If one has only a bunch of message-ids, there is no way to determine
>which articles corresponding to those message-ids are immediate
>predecessors by examining the message with those message-ids in a
>References field.
True, but why should it be necessary to know which are the immediate
predecessors?
> If all of the messages corresponding to the
>message-ids are available (not always the case) and if it is not
>inefficient to retrieve the References fields from each of those
>messages (possibly true if IMAP is used, maybe true if articles
>are directly available, probably not true if access is via NNTP),
Actually, it is rather easy in NNTP, by making use of the overview
information, which is what newsreaders usually do. There is still the
problem that some earlier messages from the thread may have expired.
>then it is possible to construct a dependency graph. However, even
>if all of those conditions are met and after the graph is constructed,
>it is still not possible to positively identify the messages being
>directly referenced in all cases (one of the predecessors may itself
>be a predecessor of another).
The prime aim of a newsreader should be to present the articles in an
order such that an article is never presented earlier than one of its
predecessors. You do not need the complete dependency tree for that,
though it may be possible to do a better job if you have it.
Presenting the articles nicely indented is fine if you can do it, but if
they form a graph rather than a tree then it is not possible to do so.
Sensible methods of doing this job ought to be investigated, but I see no
fundamental reason why they would need to assume an ability to detect
immediate predecessors. Any such scheme is unlikely to be deployed if it
needs to rely on changes to the syntax of the References header or on the
presence of other headers.
-- 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@clerew.man.ac.uk 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