On Fri, Dec 11, 2015 at 12:34:29PM -0800, Kevin J. McCarthy wrote:
> On Thu, Dec 10, 2015 at 10:12:37PM -0800, Will Yardley wrote:
> > On Thu, Dec 10, 2015 at 05:49:40PM -0800, Kevin J. McCarthy wrote:
> > > What happens with unpatched mutt-1.5.23 if you unset reflow_text?
> > 
> > However, if I'm understanding the description right, it seems like this
> > should be ignored like $indent_string (or at least turned off when not
> > explicitly set) if $text_flowed is unset
> 
> It makes sense to ignore $indent_string when $text_flowed is set because
> the indentation is set by the RFC.
> 
> $reflow_text is used when displaying (and generating reply text), which
> are somewhat independent from whether you want to generate flowed emails
> or not.

The docs say:

 reflow_text

   Type: boolean
   Default: yes

   When set, Mutt will  reformat  paragraphs  in text/plain
   parts marked format=flowed. If unset, Mutt will display
   paragraphs

So based on that, I understood this to mean that $reflow_text and
$reflow_wrap are specifically applied to text parts marked as
format=flowed text only. Am I reading the man page wrong on this?

The odd thing is that, while you say $reflow_text affects display, I
see:

>> foo
in both the raw message (if I hit 'edit-message' to view the raw message
in my editor), and in in the internal pager, but get:

>>foo
in my editor when replying to same (though my memory is that it used to
show up as >>foo in the pager too).

So I think the *display* issue was either fixed separately, or is
controlled by a different setting.

> > ... are there other conditions under which this setting would
> > be useful? I would think that sending the reply through the
> > rfc3676_handler when not generating flowed_text would be broken
> > behavior.
 
> It's certainly confusing!  But I don't think it's broken.  Reflowing
> might make it easier when replying using fixed width (ignoring the
> current quoting issue) by allowing you to control the width in the reply
> yourself (via $reflow_wrap).  Also, I think it gives us an opportunity
> to "fix" the quoting, which you argue next:

Yes, but in the case where the text isn't flowed, I think there's a
better argument that one shouldn't monkey with the 

format=flowed text has a very strict interpretation of what's "quoted"
or not (though obviously mailers that *generate* f=f may sometimes
improperly determine what's quoted when quoting material generated using
conventional quoting).

In other words, f=f
>>> foo
means "foo" with 3 quoting levels, whereas with a text/plain message
with the content:
> > > foo

we would typically assume the same. But we might also see super-cite
style quotes (FOO>), or just literal leading ">"s.


> On Thu, Dec 10, 2015 at 10:27:31PM -0800, Will Yardley wrote:
> > ps - Given text known to be f=f, I'd also prefer to see the quoted
> > section like:

> > > > > Yay!
> > > >
> > > > Noes!
> > 
> > vs.
> > 
> > > >> Yay!

> > which is more consistent (and, I think, what the $stuff_all_quoted may
> > have done). Since with format=flowed text, lines starting with a '>' are
> > quoted, and the depth is determined by the number of '>' characters, I'm
> > not sure that the argument that you're reformatting the message in an
> > unexpected way carries here
[...]
> I can understand this if we route through the text_plain_handler() -
> there we have no business messing with content.  But in rfc3676_handler,
> as you say, the quotes are conceptual.  If $text_flowed is unset, I
> don't see why we couldn't allow for the quotes to be represented
> differently.

Yes! That's kind of what I was getting at, and yes, agree we should not
mess with the former case.

Since most mutt users are *not* generating format=flowed messages (and I
still argue that, even with an editor that supports it, it's tricky to
really do well in a mailer that doesn't have its own editor), I do think
it's important to make things work optimally for this common use case
(*replying* to f=f messages when not generating them).

> A couple ideas come to mind.  One might be a boolean setting that adds
> spaces after each quote at each level.  So ">>>Foo" turns into
> "> > > Foo".  Another idea would be a setting that specifies the desired
> quote string yourself.  So, to achieve the same result, you could set it
> to "> ".

It seems to me that it should just honor $indent_string, since that's
already documented. Just, in the case of f=f text, there could be an
additional option to *enforce* indent_string on the entire quoted
section.

w

Reply via email to