bug#20663: page.el (forward-page): Avoid skipping pages

[Marcin Borkowski]

On 2016-04-11, at 15:35, Eli Zaretskii <eliz@gnu.org> wrote:

>> From: Marcin Borkowski <mbork@mbork.pl>
>> Cc: Eli Zaretskii <eliz@gnu.org>,  20663@debbugs.gnu.org
>> Date: Mon, 11 Apr 2016 12:20:07 +0200
>> 
>> I spent a few minutes on this with Edebug and Git today.
>
> Thanks.
>
>> > - All the patch does it make the code consistent with its comments, that 
>> > is:
>> >
>> >    ;; In case the page-delimiter matches the null string,
>> >    ;; don't find a match without moving.
>> 
>> Seems ok.
>> 
>> > - If you try the recipe (I just did on Emacs 24.5.1, don't have time to 
>> > check
>> > now on upstream), you'll see that a page gets skipped, which is not the 
>> > desired
>> > behaviour.
>> 
>> Condirmed.  And it doesn't happen with the patch installed.
>
> Can one of you please explain why the original code misbehaves?

Quoting from the original report:

--8<---------------cut here---------------start------------->8---
> Running `emacs -Q example.txt`:
>
>       M-<
>       C-x n p
>       M->
>       M-1 C-x n p
>
> This should bring us from page 1 to page 2, but page 3 gets displayed instead.
--8<---------------cut here---------------end--------------->8---

>> For instance, when the point is -!-, is the point on the first or
>> second page here?  Emacs with and without the patch has different
>> opinions on that.
>
> Good point.

Here's the relevant excerpt from the manual:

--8<---------------cut here---------------start------------->8---
   The variable ‘page-delimiter’ controls where pages begin.  Its value
is a regular expression that matches the beginning of a line that
separates pages (*note Regexps::).  The normal value of this variable is
‘"^\f"’, which matches a formfeed character at the beginning of a line.
--8<---------------cut here---------------end--------------->8---

Is it me or is the above ambiguous?


I'm a bit busy now, but I'll make a second attempt at this issue within
a few days.  One of the problems is that (as the above paragraph seems
to confirm) the very notion of a "page" in Emacs is vague.  IMHO we
should start with a clear definition of a "page".  It is well possible
that different functions in page.el use different interpretations of
this notion, and the bug is just a symptom if such a mess.

My proposal is that a "page separator" would be a position in the buffer
where (looking-at-p page-delimiter) is true, and if point is at such
a place, then we consider it on the next page.  I.e., in this situation

abcabcabc
-!-^L
cbacbacba

the point is already on the second page (unlike the default Emacs
behavior).

Then, someone should study page.el and where necessary, update it to the
precisely defined notion of a "page".  (The mythical "someone" might be
me.)

WDYT?

-- 
Marcin Borkowski
http://octd.wmi.amu.edu.pl/en/Marcin_Borkowski
Faculty of Mathematics and Computer Science
Adam Mickiewicz University