"Charles Lindsey" <chl(_at_)clw(_dot_)cs(_dot_)man(_dot_)ac(_dot_)uk> writes:
In <iluu1e9mz83(_dot_)fsf(_at_)latte(_dot_)josefsson(_dot_)org> Simon Josefsson
<jas(_at_)extundo(_dot_)com> writes:
Section 5.1 "Generating Format=Flowed" currently says "lines
SHOULD be shorter than 80 characters"....
2) Leave the original message as is, and potentially break the
SHOULD about lines being shorter than 80 characters.
If the original message was not format flowed, then it Ought to
have
started out with a line length of 72 or thereabouts to allow
room for
quoting. So in a well-ordered thread the situation should not
arise, and
if it does, then things are no worse than they ar at present
with unflowed
messages.
Your message illustrates quite well that this isn't always true
(your first line is 82 characters long, now it is 84...). The
longest of the lines I quote is now 76 characters, which is only a
few quote iterations away from breaking the limit. The problem is
that a format=flowed implementation has to break a SHOULD in order
to quote this message. The normative text in the specification
should be more realistic, IMHO.
But I agree that letting it expand beyond 80 characters is the
correct
thing to do if it arises.
Good, that's what I'd like to see added.
However, an alternative would be to reformat the quoted text (I
sometime
do this using par, which has nice facilities for the
purpose). This should
be under user control, though, as it could make a mess in
inappropriate
cases. However, if it is done then the format could then be made
flowed.
This is what my implemention does too, using the filling functions
in emacs, which often makes good guesses. But, and this triggered
my review of the implementation, it sometimes produces horrible
results.
So if the draft is to mention this at all, maybe that
alternative could be
mentioned also.
I agree.