From: Bless Terje (link@rito.no)
Date: Sun Aug 22 1999 - 11:41:58 CDT
Henry Spencer <henry@spsystems.net> wrote:
>Terje Bless <link@tss.no> wrote:
>
>>As long as the body is unstructured you can't very
>>well encourage people to try to parse it.
>
>You don't need to parse it to search it for a particular byte
>sequence.
No, but the body of an article is by definition an uninterpreted "blob" of
data. If you let a particular byte sequence in it alter the behaviour of the
agent you also need a way to indicate that this particular byte sequence
should /not/ alter the behaviour. In effect you've defined semantical
meaning to the contents of the body, with processing instructions and escape
codes.
Do you really want to go there with text/plain?
>However... as the originator of that particular idea, I have
>to say that it's certainly weakened by the lack of any current
>implementation. I still think it's a good idea, but the case
>for it is not strong.
I'm not really sure I see the significance of the lack of any
implementations. It's an optional feature and one that doesn't gain you very
much. It's fairly hard to implement in that it requires you to scan the body
-- something which most, in my experience, newsreaders don't do -- and so
you need to add a whole new set of code for this little feature. This
doesn't really reflect on the idea, but rather on tradeoff involved in
implementing it.
I'm not even sure I think it's a very good idea. I don't really see what you
gain by having any mentioned Message-IDs left in the References field.
Especially when you start deleting message IDs so you can't depend on any
adjacent Message-IDs being from direct precursors or followups (you lack
context).
However, the only real problem I see with it is that it doesn't, IMO, work
well in combination with text/plain message bodies. If we had a structured
message body -- one where we could unambigously say "this is a Message-ID
and not just a random sequence of bytes" -- it would work a lot better.
-link