Re: Strange problems with 'format=flowed' in Content-Type header

[Simon Josefsson <simon@josefsson.org>]

Florian Kruse <flokru@flokru.gnuu.de> writes:

> Often there are also empty lines which are not supposed to be
> there.

XEmacs `fill-region' add empty lines, Emacs `fill-region' does not.
`fill-flowed' probably need to implement filling itself...

> The signature delimiter isn't '-- ' (which it would be without
> 'format=flowed') but '--'.

Right. RFC 2646 doesn't seem to say anything about _decoding_ "-- ".
This means it should be flowed.  Not what you'd expect.  I've made a
special case for this in the code now.

> > I think we need the original article to be able to say anything much.
> 
> OK, here it is:

Thanks, I noticed two problems and they turned out to be the same:

,----
|  =20
| > Eine gute Entscheidung, denn schlechter als OE ist eigentlich nur Lotus=
| 
| > Notes.
`----

is turned into

,----
|   > Eine gute Entscheidung, denn schlechter als OE ist eigentlich nur
| Lotus
| > Notes.
`----

Could you see if CVS Gnus work better for you?

It would be good if this was reported to the Mozilla people, they are
creating invalid messages.  (Details: They are breaking the
"quote-depth win" rule in rfc2646.  The first empty line have a
quote-depth of 0 and is flowed into a line with quote-depth of 1.
That's bad.)