Re: recent message.el quote changes.

[Simon Josefsson <jas@extundo.com>]

Florian Weimer <fw@deneb.enyo.de> writes:

>> ,----
>> | >I think some servers MIME parse messages and store them in another
>> | >equivalent format somehow, and when clients requests RFC822-messages
>> | >it reconstructs MIME delimiters and MIME headers.  All the 
>> | information
>> | >is there, but since the signature was calculated over the first MIME
>> | >delimiters/headers the reconstruction often breaks the signature.
>> `----
>
> I think I'm going to write a comb quoting detector and restorer.

Maybe changing the current algorithm to assume that a line is quoted
only if preceeded by blank line into a negative; "a line is not
considered quoted if it starts with > and the next character is
[0-9]"?

I don't think words beginning with ">" are that common that we need to
assume that lines beginning with such words need to be preceeded by
blank lines to be regarded as quoted.  (Five points for anyone who
managed to follow that logic.)  I'm not sure I like the
blank-lines-preceed-quoted-lines assumption, there's no specification
I know of that suggests that.

I'm starting to lean towards my first patch, I'd rather have a simple
algorithm that fails sometimes than a complicated algorithm that fails
sometimes.  As it stands, the simple algorithm fails less frequent
than the complicated as well.

> BTW, does Outlook always use '>' and never '> '?

Maybe that approach is simpler/better, and it is OK with both usefor
and RFC2646.  But it doesn't look good.