On Sat, 28 Jan 2012 00:09:58 -0500, Austin Clements <amdragon@MIT.EDU> wrote:
> Quoth David Edmondson on Jan 25 at  1:48 pm:
> > There's no need to call `truncate-string-to-width' twice in this code
> > path.
> 
> LGTM if what I point out below is okay.  Technically this changes the
> behavior of this code, but what it did before was obviously wrong (if
> you do roll a new version, I'd mention this in the commit message; but
> no need to do that just for this).
> 
> > ---
> >  emacs/notmuch.el |   14 ++++++--------
> >  1 files changed, 6 insertions(+), 8 deletions(-)
> > 
> > diff --git a/emacs/notmuch.el b/emacs/notmuch.el
> > index 3ec0816..3f6b977 100644
> > --- a/emacs/notmuch.el
> > +++ b/emacs/notmuch.el
> > @@ -441,18 +441,16 @@ Complete list of currently available key bindings:
> >    (interactive "P")
> >    (let ((thread-id (notmuch-search-find-thread-id))
> >  	(subject (notmuch-search-find-subject)))
> > +
> > +    (if (string-match "^[ \t]*$" subject)
> > +	(setq subject "[No Subject]"))
> > +
> 
> Is subject necessarily a string at this point?  Previously this only
> ran if the code determined there was a thread at point.

No, it's a bug. It's fixed in the third patch, but should be fixed here
as well.