On Tue, 17 Feb 1998, I wrote
If the only use of text/paragraph will be to allow misusers of
text/plain to correctly label their stuff, then the current proposal is
fine...
...however, I am still worried about the effects on message-quoting if
it gets wider use. It will be all too easy for quoted sections to become
badly wrapped in long conversation threads leading to users complaining
about being mis-quoted. Perhaps an explicit warning to MUAs preparing
reply messages is needed.
The more I think about this, the more I worry about the interaction
between text/paragraph and message-quoting. I fear we may be moving from a
standard (text/plain) that's being broken to a standard (text/paragraph)
that _is_ broken.
A reply to a paragraph within a text/paragraph body part that looks like
A message with ... ... a very long line.
would, if the entire paragraph is quoted with the standard marker, be
>A message with ... ... a very long line.
which may get displayed as
>A message with ...
>... a very long line.
or
>A message with ...
... a very long line.
depending on whether the MUA decides to do anything at all about quoted
material (current draft says MUAs "MAY wish to consider" quoting
conventions).
The user, if presented with the second version, will say that the message
display is broken because text is being mis-attributed. The retort here
could simply be that the MUA displaying the message is broken because it
is displaying the message sub-optimally. Bad, but the finger of blame gets
pointed at that MUA.
If that paragraph gets requoted, it may get requoted as (using quoting
methods used in other messages in this thread):
>>A message with ... ... a very long line.
or
> >A message with ... ... a very long line.
or
> A message with ... ... a very long line.
or, using other quote characters (thankfully seen mainly on UseNet)
: >A message with ... ... a very long line.
or
= >A message with ... ... a very long line.
or any number of variations
In order to display this message properly to the user (ie to give the
correct impression of who said what), the receiving MUA is going to have
to successfully parse these forms. Any failure in display can't really be
put at the door of the MUA as there is no RFC that governs quoting
conventions - the failure comes from the media-type itself. At this point,
though, the message itself won't be broken since examining the source
line-breaks will determine correctly the authors involved.
However, as soon as the user splits paragraphs in order to quote at the
most relevant point, the display problem becomes a problem with the reply
source itself. The final recipient will then be unable to deduce the
correct authorship of the text.
The conclusion must be that using _any_ quote character when replying in
text/paragraph may cause damage to the ensuing conversation thread. The
problem does not lie in MUAs but in text/paragraph itself.
I think the draft must be changed in this area: perhaps to deprecate the
use of quote characters within text/paragraph, or maybe to say that MUAs
SHOULD downgrade to text/plain before quoting. Maybe even to deprecate
text/paragraph itself and make it clear that the RFC (to be) only exists
to deal with the current practice of misusing qp and text/plain.
--
Ian Bell T U R N P I K E Ltd