<<< Date Index >>>     <<< Thread Index >>>

Re: [PATCH] fix indent_string handling with format=flowed



Hi,

On Wed, Oct 10, 2007 at 07:01:50PM +0200, Rocco Rutte wrote:
> * Johannes Stezenbach [07-10-10 18:01:15 +0200] wrote:
>
>> changeset:   5109:d3e0d6c60d87
>> branch:      HEAD
>> user:        Brendan Cully <brendan@xxxxxxxxxx>
>> date:        Tue Apr 10 13:46:09 2007 -0700
>> summary:     Don't use $indent_string to quote replies to flowed messages 
>> (closes #2874).
>
>> introduced a regression when replying to format=flowed mails.
>
> What kind of regression exactly?

"Don't use $indent_string to quote replies to flowed messages" is the
regression.

>> (Note: It would make sense to ignore $indent_string if
>> $text_flowed is set. My patch doesn't address this as
>> I'm never creating format=flowed mails, I only have to
>> reply to them.)
>
> I didn't really try it nor did I look closely at bug #2874. But the problem 
> with using something other than '>' for quoting (regardless whether you 
> have $text_flowed set or not) is: it'll break when one someone after you 
> replies with $text_flowed set at some point in the message tree.
>
> Suppose you get a message quoted as '>' already. Now you reply, so we have 
> '|>'. When somebody with $text_flowed set replies, you'll have '>|>' as a 
> f=f message. The quoting level of 3 gets mangled to 1 only. That per se 
> wouldn't be much of a problem, but lines still have trailing spaces so 
> clients may attempt to reflow them keeping the '|>' upon concatenation.
>
> So, IMO mutt should either a) really ignore $indent_string when f=f is 
> involved or b) should allow $indent_string and strip the spaces when 
> flowing lines together.
>
> a) is what we have now, but maybe the docs for $indent_string should make 
> clear that it's not used when $text_flowed is set _or_ $text_flowed is 
> unset and one replies to f=f messages.

OK, sorry if I put you on the wrong track with the "|" thing -- this was
just used during testing. What I really want to do (and what worked before
but broke with the last mutt release) is to feed Outlook users the same
crap that they send me (they want it that way), but there are also a
few AppleMail users in the mix (they're the ones sending f=f):

(I know you'll love this...)

folder-hook . 'set indent_string="> "; set attribution="On %d, %n wrote:"'
folder-hook idiots/ 'set indent_string=""; set attribution="-----Original 
Message-----\nFrom: %f\nSent: %d\nTo: %t\nSubject: %s\n"'


I think when the code which my patch changes is executed, the f=f
message has already been converted to a non-f=f message, so it doesn't
make sense to suppress $indent_string handling at this stage. I could
be wrong, of course, but with the mails I have for testing my patch does
the right thing.


Thanks,
Johannes