Quoth Dmitry Kurochkin on Feb 19 at  9:29 pm:
> On Sun, 19 Feb 2012 12:25:41 -0500, Austin Clements <amdragon@MIT.EDU> wrote:
> > Quoth Tomi Ollila on Feb 19 at 11:43 am:
> > > On Sun, 19 Feb 2012 01:22:10 -0500, Austin Clements <amdragon@MIT.EDU> wrote:
> > > > Based on the thread at id:"20120213152858.GO27039@mit.edu" it seems
> > > > like people want show refresh to retain message state by default (I
> > > > certainly do), rather than reset it by default.  As a nice bonus, this
> > > > fixes a broken test.
> > > 
> > > Hmm
> > > 
> > > Every '=' keypress in a thread removes 'unread' tag from next unread 
> > > message :o
> > 
> > Oh dear.  It's slightly subtler than that, though.  Refreshing will
> > remove the unread tag from the first message matching the query.  (I
> > suspect you had tag:unread in your query?)  This is particularly
> > confusing for state-retaining refresh because that message might not
> > even be open.
> > 
> > > That is probably desirable feature when refreshing view without
> > > retaining state. Also, probably notmuch-show-refresh-view should
> > > be fixed not to mark next message unread when retaining state ?
> > 
> > I would argue that refresh should never have side-effects, so neither
> > case should mark anything read.  I'll send a v2.
> 
> I agree.  But this looks like a separate issue, no?  So why v2 instead
> of a separate patch/series?

I wanted to slip it in before inverting the default behavior of
refresh.