Re: [Nmh-workers] some indexing results

[Earl Hood]

On Tue, Feb 8, 2011 at 1:26 PM, Paul Vixie <address@hidden> wrote:
> format=flowed is now considered as the implicit default by some readers
> notably apple's Mail.app, because so many senders use the format without
> saying so in their mime headers.  MH needs to adapt to current conditions.

I have reservations about that.

I have a pet peeve of MUAs not providing the proper headers, and I
blame Microsoft for this historically bad behavior that has lead
others to replicate this bad behavior.

Before format=flowed existed, you had MUAs auto-wrapping
text for text/plain message and rendering things in non-fixed
width fonts.  Hence, it is hard to say if what is being seen
now is just a continuation of this, or if the underlying
data is truly format=flowed, but not labeled as such.

> this probably means some heuristic where if the text looks like nonwrapped
> paragraphs then format=flowed is presumed, unless some CLI option overrides
> this.  for example, a new show -autowrap / -noautowrap option defaulting
> to -noautowrap if !isatty(STDOUT_FILENO) and to -autowrap otherwise.

I think any such heuristic should be disabled by default.  nmh users
are not your typical email user.  I do not like software that
formats messages in a matter that is different from what MIME headers
indicate w/o my explicit knowledge.

In the example you cite, I would have -noautowrap always be the
default.

Of course, if the MIME content-type header does state format=flowed,
then auto-wrapping should occur as defined by the format=flowed
spec.  For cases when output is redirected to a non-tty device,
you may need an option that allows one to specify the max column
count so nmh would know what the wrap boundary is.

--ewh