Re: problem of released-button style mode-line

[Kenichi Handa]

Miles Bader <address@hidden> writes:
> Kenichi Handa <address@hidden> writes:
>>  > Thank you, but it seems that:
>>  >   (pos-visible-in-window-p POS nil t)
>>  > returns t even if the upper part of line at POS is
>>  > non-visible.
>>  
>>  It seems that this function can be used only for the current
>>  buffer, but the docstring is vague about this restriction.

> I don't understand what you mean by this.

> pos-visible-in-window-p doesn't use the current buffer at all, except to
> default the value of POS.

Please try this:
% emacs -q --no-site-file -g 80x40
C-h t C-x 2 C-x b RET
;; Now there are two windows; *scratch*  and TUTORIAL, and
;; you are in *scratch*.
C-x =   ;; you are at point 192
ESC : (pos-visible-in-window-p 191 (get-buffer-window "TUTORIAL") t) RET
        => t
ESC : (pos-visible-in-window-p 192 (get-buffer-window "TUTORIAL") t) RET
        => t
ESC : (pos-visible-in-window-p 193 (get-buffer-window "TUTORIAL") t) RET
        => nil
;; But, In the TUTORIAL buffer, point 193 is surely fully visible.

>>  In the code of pos_visible_p, you calls start_display but
>>  I thought that the arg W (window) should be that of the
>>  current buffer.

> What do you mean?  Buffers don't point to windows, windows point to
> buffers, so the current buffer doesn't *have* a window...

Sorry for the incorrect wording.  At the beginning of start_display,
init_iterator is called.   And, at the beginning of
init_iterator, we have this code:

  xassert (charpos < 0 || (charpos > 0 && charpos <= ZV));

This apparently indicates that CHARPOS should be a position
in the current buffer, not that of WINDOW's buffer.

> BTW, the fact that the POS and WINDOW arguments of
> pos-visible-in-window-p default separately seems slightly odd to me.
> What do people think of the following change:

If pos-visible-in-window-p should work for a window which
doesn't have the current buffer, I agree with this change.
But, currently not.

---
Ken'ichi HANDA
address@hidden