From: Dmitry Kurochkin <dmitry.kurochkin@gmail.com>
To: notmuch@notmuchmail.org
Subject: Re: [PATCH 2/2] emacs: skip forward to visible text in
	notmuch-show-message-extent
In-Reply-To: <1309667286-10357-2-git-send-email-dmitry.kurochkin@gmail.com>
References: <1309667286-10357-1-git-send-email-dmitry.kurochkin@gmail.com>
	<1309667286-10357-2-git-send-email-dmitry.kurochkin@gmail.com>
User-Agent: Notmuch/0.5-321-g41686e2 (http://notmuchmail.org) Emacs/23.3.1
	(x86_64-pc-linux-gnu)
Date: Tue, 05 Jul 2011 13:38:03 +0400
Message-ID: <87iprh5apw.fsf@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Precedence: list

On Sun,  3 Jul 2011 08:28:06 +0400, Dmitry Kurochkin <dmitry.kurochkin@gmail.com> wrote:
> The patch rewrites `notmuch-show-message-extent' to be more
> robust.  The main goal is to make it work as expected if point is
> invisible.  Besides, there are no more point movements and
> property search functions are used instead manual loops.  The
> comment regarding properties strangeness is removed since there
> is no strangeness here: property ranges (as well as overlay, and
> many others, I believe) are given as [begin, end), not [begin,
> end].

Please do not apply this patch.  I am considering a different solution:
instead of making the trailing newline invisible, make the preceding
newline invisible.  This way a message would always have a visible
newline at the end and it would prevent `beginning-of-visual-line' going
beyond the beginning of message.

FYI there is a discussion on emacs-devel [1] regarding point adjustment
inside invisible regions.

Regards,
  Dmitry

[1] http://thread.gmane.org/gmane.emacs.devel/141479

> ---
>  emacs/notmuch-show.el |   19 +++++++++++--------
>  1 files changed, 11 insertions(+), 8 deletions(-)
> 
> diff --git a/emacs/notmuch-show.el b/emacs/notmuch-show.el
> index f96743b..cf8b405 100644
> --- a/emacs/notmuch-show.el
> +++ b/emacs/notmuch-show.el
> @@ -933,15 +933,18 @@ All currently available key bindings:
>  
>  ;; Movement related functions.
>  
> -;; There's some strangeness here where a text property applied to a
> -;; region a->b is not found when point is at b. We walk backwards
> -;; until finding the property.
>  (defun notmuch-show-message-extent ()
> -  (let (r)
> -    (save-excursion
> -      (while (not (setq r (get-text-property (point) :notmuch-message-extent)))
> -	(backward-char)))
> -    r))
> +  (let ((p (point)))
> +    ;; if point is invisible, skip forward to visible text
> +    (while (invisible-p p)
> +      (setq p (next-single-char-property-change p 'invisible)))
> +    ;; if no visible text found, use the point
> +    (or p (setq p (point)))
> +    (or (get-text-property p :notmuch-message-extent)
> +	;; if there is no text property, skip to the previous message
> +	(and (setq p (previous-single-char-property-change
> +		      p :notmuch-message-extent))
> +	     (get-text-property p :notmuch-message-extent)))))
>  
>  (defun notmuch-show-message-top ()
>    (car (notmuch-show-message-extent)))
> -- 
> 1.7.5.4
>