cursor doesn't show through transparent images in emacs 22, unlike emacs 21

cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

i don't know how to get anything but the outermost pixels of the
cursor to show behind transparent images in emacs 22, while the cursor
body shows through the transparent part of images in emacs 21.

here's a fairly simple way to produce the effect:

  emacs -q

include the following text in the scratch buffer:

  (put-text-property 1 2 'display (find-image '((:type xpm :file "info.xpm"))))

evaluate the expression, and the first character in the buffer should
turn into the info question mark glyph.

put the cursor on the first character.  in emacs 22, only the
outermost pixel of the cursor shows.  in emacs 21, the cursor color
shows through everything but the question mark symbol, itself.

the info xpm images shipped with emacs 21.4 and 22.0.50 are identical.

is this a bug, or am i doing something wrong?

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

RE: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Drew Adams]

    emacs -q
    include the following text in the scratch buffer:
      (put-text-property 1 2 'display (find-image '((:type xpm
       :file "info.xpm"))))

That doesn't seem to do anything for me; it just returns nil. Do I need to
do something else? (I'm using Emacs on Windows.)

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/18/06, Drew Adams <drew.adams@oracle.com> wrote:
>     emacs -q
>     include the following text in the scratch buffer:
>       (put-text-property 1 2 'display (find-image '((:type xpm
>        :file "info.xpm"))))
>
> That doesn't seem to do anything for me; it just returns nil. Do I need to
> do something else? (I'm using Emacs on Windows.)

you need to find an image that comes with your emacs that has a
transparent background, and provide the right parameters to find-image
- the right type and path to the image off of the image-load-path.

i nearly missed your reply, by the way - i miss some emacs-devel
traffic, so it's a good idea to cc me directly if you want to be sure
to get my attention.
-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> i don't know how to get anything but the outermost pixels of the
> cursor to show behind transparent images in emacs 22, while the cursor
> body shows through the transparent part of images in emacs 21.

In emacs 22, we don't show a block cursor on images; instead we just
put a border on the image.  The rationale is that the old behaviour was
really ugly in most cases ... particularly when using a blinking cursor.

Is this something you really need?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/19/06, Kim F. Storm <storm@cua.dk> wrote:
> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
> > i don't know how to get anything but the outermost pixels of the
> > cursor to show behind transparent images in emacs 22, while the cursor
> > body shows through the transparent part of images in emacs 21.
>
> In emacs 22, we don't show a block cursor on images; instead we just
> put a border on the image.  The rationale is that the old behaviour was
> really ugly in most cases ... particularly when using a blinking cursor.
>
> Is this something you really need?

yes.  it's terrible the way it is.
-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

[-- Attachment #1: Type: text/plain, Size: 1472 bytes --]

On 8/19/06, Ken Manheimer <ken.manheimer@gmail.com> wrote:
> On 8/19/06, Kim F. Storm <storm@cua.dk> wrote:
> > "Ken Manheimer" <ken.manheimer@gmail.com> writes:
> >
> > > i don't know how to get anything but the outermost pixels of the
> > > cursor to show behind transparent images in emacs 22, while the cursor
> > > body shows through the transparent part of images in emacs 21.
> >
> > In emacs 22, we don't show a block cursor on images; instead we just
> > put a border on the image.  The rationale is that the old behaviour was
> > really ugly in most cases ... particularly when using a blinking cursor.
> >
> > Is this something you really need?
>
> yes.  it's terrible the way it is.

the two pictures show views of the same buffer, one with the cursor in
the window, the other with the cursor in another window.  they're
indistinguishable.  not good.

of course, there's more to the picture.  i didn't show how the mode
line lights up to indicate that the current buffer is active.
unfortunately, as i've determined over the month or so i was hoping to
get used to it (and maybe stumble on the solution, since i failed to
do so in the while i spent initially trying to figure it what i was
doing wrong), but i still look for the blatant - but, unfortunately,
absent - cue of the boldly lit cursor.  it's usually there, would be
just the right thing, and its absence is broken, at least for my
situation.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

[-- Attachment #2: allout-widgets_with.png --]
[-- Type: image/png, Size: 7774 bytes --]

[-- Attachment #3: allout-widgets_without.png --]
[-- Type: image/png, Size: 7300 bytes --]

[-- Attachment #4: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Emacs-devel mailing list
Emacs-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-devel

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> On 8/19/06, Ken Manheimer <ken.manheimer@gmail.com> wrote:
>> On 8/19/06, Kim F. Storm <storm@cua.dk> wrote:
>> > "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>> >
>> > > i don't know how to get anything but the outermost pixels of the
>> > > cursor to show behind transparent images in emacs 22, while the cursor
>> > > body shows through the transparent part of images in emacs 21.
>> >
>> > In emacs 22, we don't show a block cursor on images; instead we just
>> > put a border on the image.  The rationale is that the old behaviour was
>> > really ugly in most cases ... particularly when using a blinking cursor.
>> >
>> > Is this something you really need?
>>
>> yes.  it's terrible the way it is.
>
> the two pictures show views of the same buffer, one with the cursor in
> the window, the other with the cursor in another window.  they're
> indistinguishable.  not good.

Have you disabled cursor blinking?

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Jason Rumney]

Kim F. Storm wrote:
> In emacs 22, we don't show a block cursor on images; instead we just
> put a border on the image.  The rationale is that the old behaviour was
> really ugly in most cases
We used to do something different - like XOR the image with the cursor 
color or something. That was certainly ugly, but allowing the cursor 
color to show through transparent areas does not seem too ugly to me 
(though may involve lots of processing every time the cursor blinks, 
unless we cache two images for every image that is displayed).

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

Jason Rumney <jasonr@gnu.org> writes:

> Kim F. Storm wrote:
>> In emacs 22, we don't show a block cursor on images; instead we just
>> put a border on the image.  The rationale is that the old behaviour was
>> really ugly in most cases
> We used to do something different - like XOR the image with the cursor
> color or something. That was certainly ugly, but allowing the cursor
> color to show through transparent areas does not seem too ugly to me

Apparently you never had the whole window flashing at you, preferably
with green intermissions because of mouse-over.  Images can get
arbitrarily large.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    In emacs 22, we don't show a block cursor on images; instead we just
    put a border on the image.  The rationale is that the old behaviour was
    really ugly in most cases ... particularly when using a blinking cursor.

    Is this something you really need?

Unless this is so grave as to count as a bug, let's leave it alone, so
we can proceed to pretest.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    yes.  it's terrible the way it is.

What exactly are you trying to do, which comes out so terrible?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/20/06, Richard Stallman <rms@gnu.org> wrote:
>     yes.  it's terrible the way it is.
>
> What exactly are you trying to do, which comes out so terrible?

sorry - i did elaborate later that evening in a subsequent message,
including a few pictures that show what i feel is a real problem.
here's the details of the problem.

in my allout outliner extensions, the outline structure is represented
with graphics characters - icons for the connecting lines and item
bullets.  single character navigation is available when the cursor is
on the structural elements (like allout's "hotspot" navigation, but
extended), so browsing and even moving between edits is simplified by
moving to the structural icons.  with this bug, however, it's hard to
tell whether or not the cursor is in the window or another.

in my (now, a couple of months worth of) experience, highlighting of
the mode line is not sufficient because the cue of the highlighted
cursor is normally so pronounced that it eclipses all other cues.
(that's not a bad thing - the cursor is the central focus of active
editing, and this activation an appropriate piece of info for it to
convey, along with location.)  being used to the highlighted cursor
cue, it's absence is a problem - when returning to my editing session,
or from editing in separate frame, i frequently spend time distracted,
wondering whether or not the cursor is active.  you would think i'd
look at the mode line, but usually i move the cursor around.  talk
about the editor getting in the way of editing...

if you have doubts about the severity of the problem, please look at
the pictures in my previous message.  while it omits some information
that actually is present, it presents what i, for one, actually notice
while editing, and what's there is inadequate to gauge an important
detail - whether or not the cursor is active.

in my case, it would be sufficient to have a buffer-local variable
that says "highlight cursor behind transparent graphics".  that
variable could default to off or on, i don't care.  another provision
on top of that, which might be a thorough solution, would be to allow
such a setting per graphic - so that a buffer could have it on and
selected graphics could inhibit it, or vice versa.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> On 8/20/06, Richard Stallman <rms@gnu.org> wrote:
>>     yes.  it's terrible the way it is.
>>
>> What exactly are you trying to do, which comes out so terrible?
>
> sorry - i did elaborate later that evening in a subsequent message,
> including a few pictures that show what i feel is a real problem.
> here's the details of the problem.

I ask again: have you disabled cursor blinking?

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

      with this bug, however, it's hard to
    tell whether or not the cursor is in the window or another.

I see why this is a problem.

So I suppose that anything which makes the cursor visible
would solve the problem.  Right?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

>       with this bug, however, it's hard to
>     tell whether or not the cursor is in the window or another.
>
> I see why this is a problem.
>
> So I suppose that anything which makes the cursor visible
> would solve the problem.  Right?

I don't know why Ken has chosen not to reply on the list, but he told
me in private that he has disabled cursor blinking (or rather, lets it
commence only after 30 seconds).

Of course, the blinking will make a very visible difference for the
default settings.  So our worry is just for people who have tampered
with the defaults.

His proposal to use transparency is not really useful except for
special cases since
a) images are not always the size of his small icons, and so could be
vastly more obnoxious when blinking
b) images don't all have transparency.

Different ways to do this:

a) offer to make the cursor border on selected frames thicker

b) offer to use a different cursor color for selected frames


-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/21/06, Richard Stallman <rms@gnu.org> wrote:
>       with this bug, however, it's hard to
>     tell whether or not the cursor is in the window or another.
>
> I see why this is a problem.
>
> So I suppose that anything which makes the cursor visible
> would solve the problem.  Right?

well, visible specifically when the buffer containing the cursor is active.

(i did consider something like actually changing the icons when the
cursor was on them, but that seemed way too invasive and brittle.)

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/22/06, David Kastrup <dak@gnu.org> wrote:
> Richard Stallman <rms@gnu.org> writes:
>
> >       with this bug, however, it's hard to
> >     tell whether or not the cursor is in the window or another.
> >
> > I see why this is a problem.
> >
> > So I suppose that anything which makes the cursor visible
> > would solve the problem.  Right?
>
> I don't know why Ken has chosen not to reply on the list, but he told
> me in private that he has disabled cursor blinking (or rather, lets it
> commence only after 30 seconds).

it was just an oversight - i meant to reply to the list and you, but
forgot to hit "reply-to".

what i wrote:

  sorry, i didn't realize you had directed that to me.

  no, but i delay cursor bliking for 20 seconds, because i find it
  distracting, and only want it happening when imy attention has gone
  away to something else and is coming back.

  are you suggesting that the blinking would be enough of a cue for an
  active buffer?

> Of course, the blinking will make a very visible difference for the
> default settings.  So our worry is just for people who have tampered
> with the defaults.

the blink of the border is a distinct cue.  i still think the absence
of the solid cursor is misleading, in this case.  to put it in an
affirmative sense, the presence of the solid cursor would be more
clear.

> His proposal to use transparency is not really useful except for
> special cases since
> a) images are not always the size of his small icons, and so could be
> vastly more obnoxious when blinking

i understand that icons are often large, so that a solid cursor would
be distracting and garish, doubly so with blinking.

> b) images don't all have transparency.

i think it's reasonable in those cases to not expect the cursor to
show through.  that said, i wouldn't mind having a display option for
the images to xor the image colors, or something like that, when the
cursor is on them.

> Different ways to do this:
>
> a) offer to make the cursor border on selected frames thicker
>
> b) offer to use a different cursor color for selected frames

i suppose both of those would work.  from seeing the effect of the
solid cursor showing through in emacs 21, my impression is that it's
the most natural - consistent and clear - signal, in cases like my
character-sized icons, where it is suitable.

as i think i suggested in a previous message, the solution i would
prefer would be to have some way to qualify icons so that they get a
solid cursor in their transparent areas, and the default would be to
have it inactive.  eg, a qualifier for xpm, png, gif, and other
transparent-capable images, like :solid-cursor:

 `:solid-cursor-shows BOOLEAN'
       When true, the cursor will appear solid through the transparent
regions of the image
       when the cursor is situated on the image and the window is
active.  The default is for
       only the cursor border to show, whether or not the window is active.

by having the default be off, then the disruption of oversize blinking
icons will be avoided.

i am hardly acquainted with display system, so don't know whether this
is doable without undue effort.  xor for non-transparent icons (and
transparent ones, for that matter) could be a related option, if it
were doable.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> as i think i suggested in a previous message, the solution i would
> prefer would be to have some way to qualify icons so that they get a
> solid cursor in their transparent areas, and the default would be to
> have it inactive.  eg, a qualifier for xpm, png, gif, and other
> transparent-capable images, like :solid-cursor:
>
> `:solid-cursor-shows BOOLEAN'
>       When true, the cursor will appear solid through the transparent
> regions of the image
>       when the cursor is situated on the image and the window is
> active.  The default is for
>       only the cursor border to show, whether or not the window is active.
>
> by having the default be off, then the disruption of oversize blinking
> icons will be avoided.

Below is a patch which adds a new image :cursor property where you can
specify the actual type of cursor to show on the image in the selected
window (that is more flexible than your suggestion, and just as easy
to do).

Due to limitations in what cursors can be displayed for an image (bar
and hbar both gives a solid box??), it only works to set it to
box, hollow, and t (which means use whatever cursor-type is in the
buffer).


As an additional part of the patch, it only changes the cursor
from box to hollow for images which as greater than 4 times the
frame's default font size i.e. small images will behave like
normal glyphs with respect to the box cursor.


We can use either or both methods -- WDYT ?

*** xdisp.c	22 Aug 2006 23:28:12 +0200	1.1113
--- xdisp.c	24 Aug 2006 13:12:33 +0200
***************
*** 367,373 ****
  Lisp_Object Qimage;

  /* The image map types.  */
! Lisp_Object QCmap, QCpointer;
  Lisp_Object Qrect, Qcircle, Qpoly;

  /* Non-zero means print newline to stdout before next mini-buffer
--- 367,373 ----
  Lisp_Object Qimage;

  /* The image map types.  */
! Lisp_Object QCmap, QCpointer, QCcursor;
  Lisp_Object Qrect, Qcircle, Qpoly;

  /* Non-zero means print newline to stdout before next mini-buffer
***************
*** 21210,21217 ****
    if (!w->cursor_off_p)
      {
        if (glyph != NULL && glyph->type == IMAGE_GLYPH) {
! 	if (cursor_type == FILLED_BOX_CURSOR)
! 	  cursor_type = HOLLOW_BOX_CURSOR;
        }
        return cursor_type;
      }
--- 21210,21238 ----
    if (!w->cursor_off_p)
      {
        if (glyph != NULL && glyph->type == IMAGE_GLYPH) {
! 	struct image *img = IMAGE_FROM_ID (f, glyph->u.img_id);
! 	if (img != NULL && IMAGEP (img->spec))
! 	  {
! 	    /* If image has a :cursor property, use that in selected window.
! 	       If property value is `t', use current cursor_type.  */
! 	    Lisp_Object cursor_prop;
! 	    if ((cursor_prop = Fplist_get (XCDR (img->spec), QCcursor),
! 		 !NILP (cursor_prop)))
! 	      return (EQ (cursor_prop, Qt)
! 		      ? cursor_type
! 		      : get_specified_cursor_type (cursor_prop, width));
!
! 	    /* Just show border around images to avoid excessive blinking
! 	       for "large images".  */
! 	    if (cursor_type == FILLED_BOX_CURSOR
! 		&& (img->width > WINDOW_FRAME_COLUMN_WIDTH (w) * 4
! 		    || img->height > WINDOW_FRAME_LINE_HEIGHT (w) * 4))
! 	      cursor_type = HOLLOW_BOX_CURSOR;
! 	  }
! 	else
! 	  /* Just show border around images to avoid excessive blinking */
! 	  if (cursor_type == FILLED_BOX_CURSOR)
! 	    cursor_type = HOLLOW_BOX_CURSOR;
        }
        return cursor_type;
      }
***************
*** 23709,23714 ****
--- 23730,23737 ----
    staticpro (&QCmap);
    QCpointer = intern (":pointer");
    staticpro (&QCpointer);
+   QCcursor = intern (":cursor");
+   staticpro (&QCcursor);
    Qrect = intern ("rect");
    staticpro (&Qrect);
    Qcircle = intern ("circle");

--
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

storm@cua.dk (Kim F. Storm) writes:

> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
>> as i think i suggested in a previous message, the solution i would
>> prefer would be to have some way to qualify icons so that they get a
>> solid cursor in their transparent areas, and the default would be to
>> have it inactive.  eg, a qualifier for xpm, png, gif, and other
>> transparent-capable images, like :solid-cursor:
>>
>> `:solid-cursor-shows BOOLEAN'
>>       When true, the cursor will appear solid through the transparent
>> regions of the image
>>       when the cursor is situated on the image and the window is
>> active.  The default is for
>>       only the cursor border to show, whether or not the window is active.
>>
>> by having the default be off, then the disruption of oversize blinking
>> icons will be avoided.
>
> Below is a patch which adds a new image :cursor property where you can
> specify the actual type of cursor to show on the image in the selected
> window (that is more flexible than your suggestion, and just as easy
> to do).
>
> Due to limitations in what cursors can be displayed for an image (bar
> and hbar both gives a solid box??), it only works to set it to
> box, hollow, and t (which means use whatever cursor-type is in the
> buffer).
>
>
> As an additional part of the patch, it only changes the cursor
> from box to hollow for images which as greater than 4 times the
> frame's default font size i.e. small images will behave like
> normal glyphs with respect to the box cursor.
>
>
> We can use either or both methods -- WDYT ?

I don't like the hardwired factor of 4: after all, nothing like this
will be applied if you use large fonts.  I think we should try to make
characters and images behave comparably.  Having a general size
limitation variable available for block cursors, whether we are
talking images or characters, would be more consistent.

This size limitation could be specified as either float (multiple of
default font size) or integer (pixels?  Or 10ths of points?).

Of course, if an image has no transparency, we probably need to use a
box cursor, anyway.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/24/06, Kim F. Storm <storm@cua.dk> wrote:
> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
> > as i think i suggested in a previous message, the solution i would
> > prefer would be to have some way to qualify icons so that they get a
> > solid cursor in their transparent areas, and the default would be to
> > have it inactive.  eg, a qualifier for xpm, png, gif, and other
> > transparent-capable images, like :solid-cursor:
> >
> > `:solid-cursor-shows BOOLEAN'
> >       When true, the cursor will appear solid through the transparent
> > regions of the image
> >       when the cursor is situated on the image and the window is
> > active.  The default is for
> >       only the cursor border to show, whether or not the window is active.
> >
> > by having the default be off, then the disruption of oversize blinking
> > icons will be avoided.
>
> Below is a patch which adds a new image :cursor property where you can
> specify the actual type of cursor to show on the image in the selected
> window (that is more flexible than your suggestion, and just as easy
> to do).

yay!

> Due to limitations in what cursors can be displayed for an image (bar
> and hbar both gives a solid box??), it only works to set it to
> box, hollow, and t (which means use whatever cursor-type is in the
> buffer).
>
>
> As an additional part of the patch, it only changes the cursor
> from box to hollow for images which as greater than 4 times the
> frame's default font size i.e. small images will behave like
> normal glyphs with respect to the box cursor.
>
> We can use either or both methods -- WDYT ?

if i'm understanding correctly, i think either or both methods will
satisfy my needs - thanks!

i'm not clear how the two interact - will the :cursor setting only
affect glyphs larger than 4 times the frame's default font size?  or
does it mean that glyphs will have particular cursors if they either
are 4 times the size or if they have an explicit :cursor display
setting?  my concern is settled either way, but the documentation may
be complex if there's no simple answer.

i also wonder whether it will be confusing to have the option for
graphics formats that don't support transparency, but have no strong
opinions about this.

i agree w/david about making the scaling threshold a parameter.  that
could actually help with documentation by providing an artifact that's
conveys some of the explanation.  i would lean towards a scaling
factor rather than absolute pixels/points, but haven't thought that
through.

by the way, i (eagerly) applied your patch by hand, but am unable to
compile - i'm getting complaints about QCursor being undeclared:

xdisp.c: In function `get_window_cursor_type':
xdisp.c:21219: error: `QCcursor' undeclared (first use in this function)
xdisp.c:21219: error: (Each undeclared identifier is reported only once
xdisp.c:21219: error: for each function it appears in.)
xdisp.c: In function `syms_of_xdisp':
xdisp.c:23733: error: `QCcursor' undeclared (first use in this function)

did i miss something, or is there more to the patch?

ken

> *** xdisp.c     22 Aug 2006 23:28:12 +0200      1.1113
> --- xdisp.c     24 Aug 2006 13:12:33 +0200
> ***************
> *** 367,373 ****
>   Lisp_Object Qimage;
> [...]

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> i'm not clear how the two interact - will the :cursor setting only
> affect glyphs larger than 4 times the frame's default font size?  or
> does it mean that glyphs will have particular cursors if they either
> are 4 times the size or if they have an explicit :cursor display
> setting?  my concern is settled either way, but the documentation may
> be complex if there's no simple answer.

A :cursor property takes precedence over the other methods.

> i also wonder whether it will be confusing to have the option for
> graphics formats that don't support transparency, but have no strong
> opinions about this.

It is _your_ responsibility to use the proper cursor for your image :-)

> i agree w/david about making the scaling threshold a parameter.  that
> could actually help with documentation by providing an artifact that's
> conveys some of the explanation.  i would lean towards a scaling
> factor rather than absolute pixels/points, but haven't thought that
> through.

It's no big deal.

>
> by the way, i (eagerly) applied your patch by hand, but am unable to
> compile - i'm getting complaints about QCursor being undeclared:

You overlooked the first part of the patch which adds QCcursor...

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/24/06, Kim F. Storm <storm@cua.dk> wrote:
> "Ken Manheimer" <ken.manheimer@gmail.com> writes:

> > i'm not clear how the two interact - will the :cursor setting only
> > affect glyphs larger than 4 times the frame's default font size?  or
> > does it mean that glyphs will have particular cursors if they either
> > are 4 times the size or if they have an explicit :cursor display
> > setting?  my concern is settled either way, but the documentation may
> > be complex if there's no simple answer.
>
> A :cursor property takes precedence over the other methods.
>
> > i also wonder whether it will be confusing to have the option for
> > graphics formats that don't support transparency, but have no strong
> > opinions about this.
>
> It is _your_ responsibility to use the proper cursor for your image :-)

fair enough.

> > i agree w/david about making the scaling threshold a parameter.  that
> > could actually help with documentation by providing an artifact that's
> > conveys some of the explanation.  i would lean towards a scaling
> > factor rather than absolute pixels/points, but haven't thought that
> > through.
>
> It's no big deal.
>
> > by the way, i (eagerly) applied your patch by hand, but am unable to
> > compile - i'm getting complaints about QCursor being undeclared:
>
> You overlooked the first part of the patch which adds QCcursor...

whoops.  thanks, that did it!  i now get a full box cursor showing
through the transparent parts of my glyphs - with no additional
changes on my part.

in case some exercising of the :cursor part of the change would be
helpful, i experimented with changing the image cursor to hbar.  i was
unable to see any change - the solid cursor still shows behind the
image. i tried using text-properties-at to check whether the :cursor
property was there, and got:

(display (image :type xpm :cursor hbar :file
"/home/klm/lib/emacs/allout_dev/allout-icons-dark-bg/closed.xpm"
:ascent center :mask (heuristic t)))

i would expect to see an hbar.  my expectation or actions may be
mistaken, and i don't need to clear this up (the full box is what i
need), but can experiment further if it would be helpful.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> i would expect to see an hbar.  my expectation or actions may be
> mistaken, and i don't need to clear this up (the full box is what i
> need), but can experiment further if it would be helpful.

As I said in a previous posting, a bar or hbar cursor is shown as a
box on an image.  That's a different bug.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/24/06, Kim F. Storm <storm@cua.dk> wrote:
> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
> > i would expect to see an hbar.  my expectation or actions may be
> > mistaken, and i don't need to clear this up (the full box is what i
> > need), but can experiment further if it would be helpful.
>
> As I said in a previous posting, a bar or hbar cursor is shown as a
> box on an image.  That's a different bug.

i missed that.

hollow works as it ought, so i can confirm the :cursor part of your patch.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

David Kastrup <dak@gnu.org> writes:

> I don't like the hardwired factor of 4: after all, nothing like this
> will be applied if you use large fonts.  I think we should try to make
> characters and images behave comparably.  Having a general size
> limitation variable available for block cursors, whether we are
> talking images or characters, would be more consistent.
>
> This size limitation could be specified as either float (multiple of
> default font size) or integer (pixels?  Or 10ths of points?).

What about the simpler approach in the patch below?  

It simply uses a hollow cursor if the image size is > WxH where WxH is
max(32, default width|height), i.e. block cursor is used on all images
<= 32x32 and possibly larger images if the default font size is larger
than that.

It is not customizable, but I suppose it will give satifying results
for most practical uses.

>
> Of course, if an image has no transparency, we probably need to use a
> box cursor, anyway.

Did you mean hollow cursor?


*** xdisp.c	24 Aug 2006 22:38:58 +0200	1.1114
--- xdisp.c	25 Aug 2006 15:38:42 +0200	
***************
*** 21213,21221 ****
    /* Use normal cursor if not blinked off.  */
    if (!w->cursor_off_p)
      {
!       if (glyph != NULL && glyph->type == IMAGE_GLYPH) {
! 	if (cursor_type == FILLED_BOX_CURSOR)
! 	  cursor_type = HOLLOW_BOX_CURSOR;
        }
        return cursor_type;
      }
--- 21213,21233 ----
    /* Use normal cursor if not blinked off.  */
    if (!w->cursor_off_p)
      {
!       if (glyph != NULL && glyph->type == IMAGE_GLYPH
! 	  && cursor_type == FILLED_BOX_CURSOR) {
! 	/* Using a block cursor on large images can be very annoying.
! 	   So use a hollow cursor for "large" images.  */
! 	struct image *img = IMAGE_FROM_ID (f, glyph->u.img_id);
! 	if (img != NULL && IMAGEP (img->spec))
! 	  {
! 	    /* Arbitrarily, interpret "Large" as >32x32 and >NxN
! 	       where N = size of default frame font size.
! 	       This should cover most of the "tiny" icons people may use.  */
! 
! 	    if (img->width > max (32, WINDOW_FRAME_COLUMN_WIDTH (w))
! 		|| img->height > max (32, WINDOW_FRAME_LINE_HEIGHT (w)))
! 	      cursor_type = HOLLOW_BOX_CURSOR;
! 	  }
        }
        return cursor_type;
      }


-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

storm@cua.dk (Kim F. Storm) writes:

> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
>> i would expect to see an hbar.  my expectation or actions may be
>> mistaken, and i don't need to clear this up (the full box is what i
>> need), but can experiment further if it would be helpful.
>
> As I said in a previous posting, a bar or hbar cursor is shown as a
> box on an image.  That's a different bug.

So rather that trying to fix that (non-trivial as it must be done on
all platforms), I'd prefer to postpone installing the :cursor property
change (in some form) until after the release.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/25/06, Kim F. Storm <storm@cua.dk> wrote:
> storm@cua.dk (Kim F. Storm) writes:
>
> > "Ken Manheimer" <ken.manheimer@gmail.com> writes:
> >
> >> i would expect to see an hbar.  my expectation or actions may be
> >> mistaken, and i don't need to clear this up (the full box is what i
> >> need), but can experiment further if it would be helpful.
> >
> > As I said in a previous posting, a bar or hbar cursor is shown as a
> > box on an image.  That's a different bug.
>
> So rather that trying to fix that (non-trivial as it must be done on
> all platforms), I'd prefer to postpone installing the :cursor property
> change (in some form) until after the release.

just to be sure i'm being clear - i don't have a concern about the
hbar/box problem.  i was just exercising more of your patch because
you had gone to the trouble of implementing it.

in fact, my concerns would be settled with the changes, currently
being discussed, that enable the filled-in box cursor for glyphs of
size comparable to the frame's default font.  i hope that one of those
alternatives will be included with the release, independent of the
:cursor provisions.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > as i think i suggested in a previous message, the solution i would
    > prefer would be to have some way to qualify icons so that they get a
    > solid cursor in their transparent areas, and the default would be to
    > have it inactive.  eg, a qualifier for xpm, png, gif, and other
    > transparent-capable images, like :solid-cursor:
    >
    > `:solid-cursor-shows BOOLEAN'
    >       When true, the cursor will appear solid through the transparent
    > regions of the image

This does not make sense without more context.  For instance, what
behavior would a value of nil specify?  And what happens in
non-transparent regions of the image?

Meanwhile, something about this wording bothers me: the idea that the
cursor appears "through" part of the image.  Is this based on the
assumption that the image is "on top of" the cursor?

If so, I think that assumption is backwards and must be changed.  The
cursor is "on top of" the image; if we want to display a cursor on the
image, the image should not be allowed to block it out!

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

> Meanwhile, something about this wording bothers me: the idea that the
> cursor appears "through" part of the image.  Is this based on the
> assumption that the image is "on top of" the cursor?

A block cursor on an image is shown by using the cursor color
as the background color in the image.  That's not changed from 
emacs 21, and I don't see any reason to change it now for 22.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/25/06, Richard Stallman <rms@gnu.org> wrote:
>     > as i think i suggested in a previous message, the solution i would
>     > prefer would be to have some way to qualify icons so that they get a
>     > solid cursor in their transparent areas, and the default would be to
>     > have it inactive.  eg, a qualifier for xpm, png, gif, and other
>     > transparent-capable images, like :solid-cursor:
>     >
>     > `:solid-cursor-shows BOOLEAN'
>     >       When true, the cursor will appear solid through the transparent
>     > regions of the image
>
> This does not make sense without more context.  For instance, what
> behavior would a value of nil specify?  And what happens in
> non-transparent regions of the image?

i was using it only to elaborate certain details of my suggestion, not
as the real documentation.  as such, it was meant specifically for
those following the context.

non-transparent regions seems like a complicated issue.  i expect that
in many cases it's neither desirable nor practial to mess with them.
some systems xor the colors under mouse icons, for instance, but
there's no reason to expect the chosen colors would be particularly
harmonious, or the icon to even be recognizable with such treatment.
however, there's a bind for icons with no transparency - hopefully the
highlighted bounding box cursor is sufficient there.

> Meanwhile, something about this wording bothers me: the idea that the
> cursor appears "through" part of the image.  Is this based on the
> assumption that the image is "on top of" the cursor?

i could as easily have said that the cursor shows "in" the transparent
parts of the image, leaving the opaque parts undisturbed.

> If so, I think that assumption is backwards and must be changed.  The
> cursor is "on top of" the image; if we want to display a cursor on the
> image, the image should not be allowed to block it out!

i think we're on pretty safe ground doing having the opaque parts o
the image block out the cursor.

consider text characters - cursors being on the character could simply
block them from view, reinforcing the paradigm of being on top of the
character.  that's a terrible loss of information - generally, the
character is allowed to show through.  the character might be shown in
the inverse tone of the cursor, or some chosen color, or something.
(emacs goes with a chosen color, i guess.)

the situation with glyphs is not so simple, because they can have
many, arbitrary colors.  in my particular case, having the cursor show
in the transparent region has the outline-structure bullet icons and
guide lines acting very like elaborate text characters, except that
their intricately colored opaque parts are not changed.  it works
well.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    in fact, my concerns would be settled with the changes, currently
    being discussed, that enable the filled-in box cursor for glyphs of
    size comparable to the frame's default font.

Before considering proposals for new features to address this problem,
I want to understand the nature of the problem.  I sent mail yesterday
explaining a confusion that I think we need to clear up, in order to
make sure we understand the situation.  When this is cleared up, we
will see how to proceed.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    A block cursor on an image is shown by using the cursor color
    as the background color in the image.

I think this means that if the image blocks out the background color,
then the cursor won't appear.  Is that correct?  I think that is a problem.

But it looks like the problem that was reported is different.  How
do they differ?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    non-transparent regions seems like a complicated issue.  i expect that
    in many cases it's neither desirable nor practial to mess with them.

It is not clear to me what sort of actions the term "mess with them"
would cover.  What I am sure of is that the cursor must be visible
when it appears over a non-transparent image.

Currently, when the cursor appears over a non-transparent image,
is it visible?  If so, how does it appear?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     A block cursor on an image is shown by using the cursor color
>     as the background color in the image.
>
> I think this means that if the image blocks out the background color,
> then the cursor won't appear.  Is that correct?  I think that is a problem.

Maybe -- but it's not a new problem.  It's the same as it was in 21.x, and
I don't recall any reports of that behaviour being a problem.

> But it looks like the problem that was reported is different.  How
> do they differ?

The current complaint is that we unconditionally replace the box
cursor by a hollow cursor on images -- even small images (comparable
in size to a character glyph) where the box cursor would do fine (and
did fine in emacs 21.x).

My change addresses this problem by making a distinction between
small and "larger" images.  IMO, this is the right fix for emacs 22.

OTOH, my change does not try to address (the never-repoprted, non-existent?)
problem with using a box cursor on images.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > I think this means that if the image blocks out the background color,
    > then the cursor won't appear.  Is that correct?  I think that is a problem.

    Maybe -- but it's not a new problem.  It's the same as it was in 21.x, and
    I don't recall any reports of that behaviour being a problem.

How could it possibly not be a problem?  Is it that nobody uses
non-transparent images?  Nobody cares about seeing the cursor in buffers
that use non-transparent images?

    The current complaint is that we unconditionally replace the box
    cursor by a hollow cursor on images -- even small images (comparable
    in size to a character glyph) where the box cursor would do fine (and
    did fine in emacs 21.x).

    My change addresses this problem by making a distinction between
    small and "larger" images.  IMO, this is the right fix for emacs 22.

It is ok with me.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/27/06, Richard Stallman <rms@gnu.org> wrote:
>     > I think this means that if the image blocks out the background color,
>     > then the cursor won't appear.  Is that correct?  I think that is a problem.
>
>     Maybe -- but it's not a new problem.  It's the same as it was in 21.x, and
>     I don't recall any reports of that behaviour being a problem.
>
> How could it possibly not be a problem?  Is it that nobody uses
> non-transparent images?  Nobody cares about seeing the cursor in buffers
> that use non-transparent images?

i just experimented, filling in the transparent part of one of my
icons with a solid color.  the cursor still shows in that region,
instead of that color.

naturally figuring i made a mistake (as those of you reading this
probably do), i looked at the image with something else, and the solid
color showed where i put it.  i started twiddling with the image and,
doing an (clear-image-cache) after each change, discovered that if i
made the top-left and bottom-right corner pixels transparent, the
solid color showed as it should.  putting the solid color in one or
the other of those corners, however, causes emacs to treat it as
transparent, at least as far as showing the cursor in it.

(a bit more experimenting shows that the combinations of corners to
qualify a solid corner for treatment as transparent is a bit more
elaborate.  i'd like to see the algorithm, but satisfying that
curiousity is not sufficient reason to get acquainted with the display
system.  this behavior certainly doesn't satisfy the principle of
least surprise, and deserves some documentation, if it doesn't already
have it.)

the upshot is that the small glyphs will benefit from  kim's change,
even if they lack transparency, with the cursor color showing instead
of the color that occupies some combination of the corners -  or maybe
some other criteria, who knows.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Juri Linkov]

>     > I think this means that if the image blocks out the background color,
>     > then the cursor won't appear.  Is that correct?  I think that is a problem.
>
>     Maybe -- but it's not a new problem.  It's the same as it was in 21.x, and
>     I don't recall any reports of that behaviour being a problem.
>
> How could it possibly not be a problem?  Is it that nobody uses
> non-transparent images?  Nobody cares about seeing the cursor in buffers
> that use non-transparent images?

I guess this is not a problem because usually a border is visible around
non-transparent images with the cursor.

What a real problem is that displaying different types of cursors on small
transparent images is illogical now.  This is a serious UI problem.
Currently, the box cursor is displayed as a hollow cursor on images, but
the bar cursor is displayed as a filled box cursor.  But this should be
vice versa: the box cursor should be displayed as a box cursor that fills
transparent areas and has the 1-pixel border, but the bar cursor should be
displayed as a hollow cursor.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> the upshot is that the small glyphs will benefit from  kim's change,
> even if they lack transparency, with the cursor color showing instead
> of the color that occupies some combination of the corners -  or maybe
> some other criteria, who knows.

Emacs looks at the corners of the image to _guess_ the image's
background/tranparent color if the image format doesn't explicit
contain this information.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     > I think this means that if the image blocks out the background color,
>     > then the cursor won't appear.  Is that correct?  I think that is a problem.
>
>     Maybe -- but it's not a new problem.  It's the same as it was in 21.x, and
>     I don't recall any reports of that behaviour being a problem.
>
> How could it possibly not be a problem?  Is it that nobody uses
> non-transparent images?  Nobody cares about seeing the cursor in buffers
> that use non-transparent images?

Cursor display does not require tranparent images -- it replaces the image
background (guesses color if not known) with the cursor color.

>     My change addresses this problem by making a distinction between
>     small and "larger" images.  IMO, this is the right fix for emacs 22.
>
> It is ok with me.

Installed.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Juri Linkov <juri@jurta.org> writes:

> What a real problem is that displaying different types of cursors on small
> transparent images is illogical now.  This is a serious UI problem.
> Currently, the box cursor is displayed as a hollow cursor on images, but
> the bar cursor is displayed as a filled box cursor.  But this should be
> vice versa: the box cursor should be displayed as a box cursor that fills
> transparent areas and has the 1-pixel border, but the bar cursor should be
> displayed as a hollow cursor.

I didn't think about it that way, but you are right.
So I have installed a change to fix this.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Juri Linkov]

> I didn't think about it that way, but you are right.
> So I have installed a change to fix this.

Thank you.  Your change makes more sense.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > How could it possibly not be a problem?  Is it that nobody uses
    > non-transparent images?  Nobody cares about seeing the cursor in buffers
    > that use non-transparent images?

    I guess this is not a problem because usually a border is visible around
    non-transparent images with the cursor.

How does the border change the outcome?  Does the cursor appear in the
border?

Why is there usually a border?  Is it because Lisp programs that
display a non-transparent image usually request a border?  Is there
something about these images that causes Emacs to add a border?

In what circumstances would there not be a border?

      But this should be
    vice versa: the box cursor should be displayed as a box cursor that fills
    transparent areas and has the 1-pixel border,

I do not understand.  What does it mean to say a cursor "has the
1-pixel border"?  Which border is that, and why does the cursor
have it?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > How could it possibly not be a problem?  Is it that nobody uses
    > non-transparent images?  Nobody cares about seeing the cursor in buffers
    > that use non-transparent images?

    Cursor display does not require tranparent images -- it replaces the image
    background (guesses color if not known) with the cursor color.

Each thing that people say about this seems to conflict with what
has been said before.  I am completely lost now.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    Emacs looks at the corners of the image to _guess_ the image's
    background/tranparent color if the image format doesn't explicit
    contain this information.

Does this mean that every image should have the cursor showing through
in some places?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/28/06, Richard Stallman <rms@gnu.org> wrote:
>     Emacs looks at the corners of the image to _guess_ the image's
>     background/tranparent color if the image format doesn't explicit
>     contain this information.
>
> Does this mean that every image should have the cursor showing through
> in some places?

yes.  my experiments and kim's analysis confirm that.  juri unravelled
another part of the problem, which kim incorporated in his fix, and i
think that all the expressed concerns have been satisfied.   the world
is a better place.-)  thanks, kim!

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     Emacs looks at the corners of the image to _guess_ the image's
>     background/tranparent color if the image format doesn't explicit
>     contain this information.
>
> Does this mean that every image should have the cursor showing through
> in some places?

For all practical purposes, yes.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Juri Linkov]

>     I guess this is not a problem because usually a border is visible around
>     non-transparent images with the cursor.
>
> How does the border change the outcome?

It is possible to visually distinguish images with the border
from images without it.

> Does the cursor appear in the border?

The hollow cursor is displayed on the image as the 1-pixel border around it.

> Why is there usually a border?  Is it because Lisp programs that
> display a non-transparent image usually request a border?  Is there
> something about these images that causes Emacs to add a border?

When Lisp programs request a border, then the margin explicitly specified
by the `:margin' keyword affects how the cursor in displayed on images.
After Kim's changes, the box cursor fills the margin with the cursor's
color, and the bar cursor doesn't fill the margin and uses the hollow
cursor with the 1-pixel border.  I think this is good behavior.

> In what circumstances would there not be a border?

The only problem I see now with Kim's changes is that the box cursor
on non-transparent images with no margin is not visible at all.

E.g. try to put the box cursor on this image:

(insert-image (create-image (make-string 8 0)
			    'xbm t :width 8 :height 8
			    :background "grey75"
			    :foreground "black"
			    :relief -2
			    :margin 0
			    :ascent 'center))

The box cursor is not visible.  Maybe it should be displayed as the hollow
cursor on non-transparent images?

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Juri Linkov <juri@jurta.org> writes:

> E.g. try to put the box cursor on this image:
>
> (insert-image (create-image (make-string 8 0)
> 			    'xbm t :width 8 :height 8
> 			    :background "grey75"
> 			    :foreground "black"
> 			    :relief -2
> 			    :margin 0
> 			    :ascent 'center))
>
> The box cursor is not visible.

But there is still a visible difference when the cursor is on the image.

As I said, I believe that the current behaviour is ok for all
practical purposes...

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > Does this mean that every image should have the cursor showing through
    > in some places?

    yes.  my experiments and kim's analysis confirm that.  juri unravelled
    another part of the problem, which kim incorporated in his fix, and i
    think that all the expressed concerns have been satisfied.   the world
    is a better place.-)  thanks, kim!

Thanks to everyone who helped fix this.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    When Lisp programs request a border, then the margin explicitly specified
    by the `:margin' keyword affects how the cursor in displayed on images.
    After Kim's changes, the box cursor fills the margin with the cursor's
    color, and the bar cursor doesn't fill the margin and uses the hollow
    cursor with the 1-pixel border.  I think this is good behavior.

It sounds plausible; if you find it easy to see the cursor this way
then I expect it is good.

    The only problem I see now with Kim's changes is that the box cursor
    on non-transparent images with no margin is not visible at all.

If that is so, how about changing the cursor to a different type when
that case occurs?  It should be easy and will solve this problem.

Kim wrote:

    But there is still a visible difference when the cursor is on the image.

What is that visible difference?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Juri Linkov]

>     The only problem I see now with Kim's changes is that the box cursor
>     on non-transparent images with no margin is not visible at all.
>
> If that is so, how about changing the cursor to a different type when
> that case occurs?  It should be easy and will solve this problem.

Changing the cursor to the hollow cursor on non-transparent images
will solve this problem.

> Kim wrote:
>
>     But there is still a visible difference when the cursor is on the image.
>
> What is that visible difference?

I see no visible difference.  Kim?

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/28/06, Kim F. Storm <storm@cua.dk> wrote:
> Juri Linkov <juri@jurta.org> writes:
>
> > E.g. try to put the box cursor on this image:
> >
> > (insert-image (create-image (make-string 8 0)
> >                           'xbm t :width 8 :height 8
> >                           :background "grey75"
> >                           :foreground "black"
> >                           :relief -2
> >                           :margin 0
> >                           :ascent 'center))
> >
> > The box cursor is not visible.
>
> But there is still a visible difference when the cursor is on the image.

to which juri responded:

< I see no visible difference.  Kim?

i also see no visible difference when i do or don't put the cursor on
the image - certainly no sign of the cursor, when it's there.  (was
the corners heuristic defeated by this particular image?)

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

"Ken Manheimer" <ken.manheimer@gmail.com> writes:

> On 8/28/06, Kim F. Storm <storm@cua.dk> wrote:
>> Juri Linkov <juri@jurta.org> writes:
>>
>> > E.g. try to put the box cursor on this image:
>> >
>> > (insert-image (create-image (make-string 8 0)
>> >                           'xbm t :width 8 :height 8
>> >                           :background "grey75"
>> >                           :foreground "black"
>> >                           :relief -2
>> >                           :margin 0
>> >                           :ascent 'center))
>> >
>> > The box cursor is not visible.
>>
>> But there is still a visible difference when the cursor is on the image.
>
> to which juri responded:
>
> < I see no visible difference.  Kim?
>
> i also see no visible difference when i do or don't put the cursor on
> the image - certainly no sign of the cursor, when it's there.  (was
> the corners heuristic defeated by this particular image?)

I don't think it is applied unless one uses the :heuristik-mask
property or its equivalent.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/29/06, David Kastrup <dak@gnu.org> wrote:
> "Ken Manheimer" <ken.manheimer@gmail.com> writes:
>
> > On 8/28/06, Kim F. Storm <storm@cua.dk> wrote:
> >> Juri Linkov <juri@jurta.org> writes:
> >>
> >> > E.g. try to put the box cursor on this image:
> >> >
> >> > (insert-image (create-image (make-string 8 0)
> >> >                           'xbm t :width 8 :height 8
> >> >                           :background "grey75"
> >> >                           :foreground "black"
> >> >                           :relief -2
> >> >                           :margin 0
> >> >                           :ascent 'center))

> >> > The box cursor is not visible.
> >>
> >> But there is still a visible difference when the cursor is on the image.
> >
> > to which juri responded:
> >
> > < I see no visible difference.  Kim?
> >
> > i also see no visible difference when i do or don't put the cursor on
> > the image - certainly no sign of the cursor, when it's there.  (was
> > the corners heuristic defeated by this particular image?)
>
> I don't think it is applied unless one uses the :heuristik-mask
> property or its equivalent.

looks like that's the interface i was missing, for controlling how the
cursor is displayed when on an image.  the info for it even mentions
that the default heuristic is to assume the background color is the
one that occurs most frequently in the images four corners.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Juri Linkov <juri@jurta.org> writes:

>>     The only problem I see now with Kim's changes is that the box cursor
>>     on non-transparent images with no margin is not visible at all.
>>
>> If that is so, how about changing the cursor to a different type when
>> that case occurs?  It should be easy and will solve this problem.
>
> Changing the cursor to the hollow cursor on non-transparent images
> will solve this problem.

We already have this code to deal with cursor on non-transparent
images:

	      /* ...        If the image doesn't have a mask, make
	     a block cursor visible by drawing a rectangle around
	     the image.  I believe it's looking better if we do
	     nothing here for mouse-face.  */
	  if (s->hl == DRAW_CURSOR)
	    {
	      int r = s->img->relief;
	      if (r < 0) r = -r;
	      XDrawRectangle (s->display, s->window, s->gc, x - r, y - r,
			      s->slice.width + r*2 - 1,
			      s->slice.height + r*2 - 1);
	    }

The problem is that if there is no border (i.e. relief), it draws nothing.

One fix would be to draw into the image.

>
>> Kim wrote:
>>
>>     But there is still a visible difference when the cursor is on the image.
>>
>> What is that visible difference?
>
> I see no visible difference.  Kim?

I guess it is because I use a font size which is bigger than the image
-- then I see the cursor in the background not occupied by the image.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > > i also see no visible difference when i do or don't put the cursor on
    > > the image - certainly no sign of the cursor, when it's there.  (was
    > > the corners heuristic defeated by this particular image?)
    >
    > I don't think it is applied unless one uses the :heuristik-mask
    > property or its equivalent.

    looks like that's the interface i was missing, for controlling how the
    cursor is displayed when on an image.  the info for it even mentions
    that the default heuristic is to assume the background color is the
    one that occurs most frequently in the images four corners.

If I understand these paragraphs right, the first one
seems to say that the cursor does not appear, and the second
one says the cursor will not appear unless the programmer has used
the :heuristic-mask property.  Did I understand them right?

If that is true, we have a bug, right?

But the third paragraph seems to say that the default for
:heuristic-mask is such as to make sure the cursor appears.
Did I understand that right?

If that is true, then maybe we don't have a bug.

What's the real situation here?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/30/06, Richard Stallman <rms@gnu.org> wrote:
>     > > i also see no visible difference when i do or don't put the cursor on
>     > > the image - certainly no sign of the cursor, when it's there.  (was
>     > > the corners heuristic defeated by this particular image?)
>     >
>     > I don't think it is applied unless one uses the :heuristik-mask
>     > property or its equivalent.
>
>     looks like that's the interface i was missing, for controlling how the
>     cursor is displayed when on an image.  the info for it even mentions
>     that the default heuristic is to assume the background color is the
>     one that occurs most frequently in the images four corners.
>
> If I understand these paragraphs right, the first one
> seems to say that the cursor does not appear, and the second
> one says the cursor will not appear unless the programmer has used
> the :heuristic-mask property.  Did I understand them right?
>
> If that is true, we have a bug, right?
>
> But the third paragraph seems to say that the default for
> :heuristic-mask is such as to make sure the cursor appears.
> Did I understand that right?
>
> If that is true, then maybe we don't have a bug.
>
> What's the real situation here?

when i tried it, the cursor wouldn't show unless i included a ":mask
'heuristic" property.  by default, i was referring to the type of
heuristic used when you specify 'heuristic without a specific type.

if i understand correctly from kim and juri's discussion, the only
reason kim saw the cursor in his solid image without specifying a mask
was because he use an image smaller than his default character size,
so there was some empty border around the image.

i believe kim suggested something like showing a margin with the
cursor color around such images when there is no such border, which i
expect is a good for that remaining case.
-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    The problem is that if there is no border (i.e. relief), it draws nothing.

    One fix would be to draw into the image.

I think that is better than nothing; if we don't get a better idea,
please implement that one.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    when i tried it, the cursor wouldn't show unless i included a ":mask
    'heuristic" property.  by default, i was referring to the type of
    heuristic used when you specify 'heuristic without a specific type.

I still am not sure I understand you.  Are you saying that when you
specify `heuristic' for :mask, then the default is to choose one of
the four corners?

What happens when you do not specify any :mask attribute?
Is that the case where the cursor fails to show?

If so, then I understand; but until you confirm, I am not sure I
understand.

If I understand right, then this is a bug, as I previously said.

    i believe kim suggested something like showing a margin with the
    cursor color around such images when there is no such border, which i
    expect is a good for that remaining case.

That would be bad,  since it would change the geometry.
What I think he proposed is to show the cursor over the outer parts
of the image.  That seems good for this case.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 8/31/06, Richard Stallman <rms@gnu.org> wrote:
>     when i tried it, the cursor wouldn't show unless i included a ":mask
>     'heuristic" property.  by default, i was referring to the type of
>     heuristic used when you specify 'heuristic without a specific type.
>
> I still am not sure I understand you.  Are you saying that when you
> specify `heuristic' for :mask, then the default is to choose one of
> the four corners?

something like that.  (it does something complicated with combinations
of corners to choose the color to use as background.)

> What happens when you do not specify any :mask attribute?
> Is that the case where the cursor fails to show?

yes.

> If so, then I understand; but until you confirm, I am not sure I
> understand.
>
> If I understand right, then this is a bug, as I previously said.

i think i agree.  the only question i have concerns the resonsibility
and disrcetion of the application developer.

re responsiblity - if application developers are given some recourse
to deliberately enable cursor visibility on their images, is it the
right thing to have enabling it be the default?  i'm inclined to think
so, but am not certain.  (my reason for raising the issue in the first
place was because i had not found any such recourse, and needed it.)

the discretion issue concerns whether it should be possible for the
developer to disable visibility of the cursor in images of their
choice.  i feel pretty clear that should be possible, and probably
should assume that was going to be provided, but want to make sure
that's factored into any solution.

>     i believe kim suggested something like showing a margin with the
>     cursor color around such images when there is no such border, which i
>     expect is a good for that remaining case.
>
> That would be bad,  since it would change the geometry.
> What I think he proposed is to show the cursor over the outer parts
> of the image.  That seems good for this case.

i agree.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

(no subject)

[Richard Stallman]

In-reply-to: <2cd46e7f0609010846o7b393dd5xef60a645125d08@mail.gmail.com>
	(ken.manheimer@gmail.com)
Subject: Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21
bcc: rms-outgoing@gnu.org
Reply-to: rms@gnu.org
References: <2cd46e7f0608181622w23c7d2b0h9b963deeadfd1c06@mail.gmail.com>
	 <E1GHpIm-00027E-7t@fencepost.gnu.org> <87wt8sbhux.fsf@jurta.org>
	 <m3zmdoa2h7.fsf@kfs-l.imdomain.dk>
	 <2cd46e7f0608291431k1d26dac9l1f41c271da28495e@mail.gmail.com>
	 <85y7t7mdha.fsf@lola.goethe.zz>
	 <2cd46e7f0608291512ne9a865av7fb7bb537658887a@mail.gmail.com>
	 <E1GIUKm-0000Gd-Tx@fencepost.gnu.org>
	 <2cd46e7f0608301120q190d710bya04f85359e79e29@mail.gmail.com>
	 <E1GIr5F-0000KL-6z@fencepost.gnu.org> <2cd46e7f0609010846o7b393dd5xef60a645125d08@mail.gmail.com>
--text follows this line--
    > What happens when you do not specify any :mask attribute?
    > Is that the case where the cursor fails to show?

    yes.

    > If I understand right, then this is a bug, as I previously said.

    i think i agree.  the only question i have concerns the resonsibility
    and disrcetion of the application developer.

I think we have a responsibility to make sure that the default way of
displaying an image does work properly with the cursor.  To have some
peculiar option which means the cursor does not display properly might
be excusable, but it's not excusable for the default to display wrong.

I suggested:

    That would be bad,  since it would change the geometry.
    What I think he proposed is to show the cursor over the outer parts
    of the image.  That seems good for this case.

Is anyone working on this?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

> I think we have a responsibility to make sure that the default way of
> displaying an image does work properly with the cursor.  To have some
> peculiar option which means the cursor does not display properly might
> be excusable, but it's not excusable for the default to display wrong.
>

But it's not a new default -- it's been like that since 21.1.

Nobody has complained about it until now, so it's not a big problem
and there is a trivial workaround.

> I suggested:
>
>     That would be bad,  since it would change the geometry.
>     What I think he proposed is to show the cursor over the outer parts
>     of the image.  That seems good for this case.
>
> Is anyone working on this?

I don't think so.  

PS: Changing this requires updates to all of X, W32 and MAC ports.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > I think we have a responsibility to make sure that the default way of
    > displaying an image does work properly with the cursor.  To have some
    > peculiar option which means the cursor does not display properly might
    > be excusable, but it's not excusable for the default to display wrong.
    >

    But it's not a new default -- it's been like that since 21.1.

Maybe that just means everybody specifies :mask.
This is definitely a bug, and I think we should fix it.

    and there is a trivial workaround.

What exactly is the workaround you recommend?

    PS: Changing this requires updates to all of X, W32 and MAC ports.

That depends on how it is done.  Let's see how that workaround works.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     But it's not a new default -- it's been like that since 21.1.
>
> Maybe that just means everybody specifies :mask.

Or that nobody uses images without a border.

> This is definitely a bug, and I think we should fix it.
>
>     and there is a trivial workaround.
>
> What exactly is the workaround you recommend?

Use :mask ...

>
>     PS: Changing this requires updates to all of X, W32 and MAC ports.
>
> That depends on how it is done.

Writing a border into the image requires those updates.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    > What exactly is the workaround you recommend?

    Use :mask ...

Then how about changing the default value for :mask?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

>     > What exactly is the workaround you recommend?
>
>     Use :mask ...
>
> Then how about changing the default value for :mask?

There is no sensible default.  What is useful as a mask depends on the
image in question.  Using heuristic-mask can mean that the whole image
flashes, but it can also mean that just a single corner pixel blinks.

Even when a mask is specified, using the whole mask as a cursor
indication is not a good idea, since a large image with a background
will then flash horribly.  The right fix in my opinion would be to use
the border of the image (extending into the image itself if the border
is too small, overwriting the image when no mask is given, but maybe
heeding the mask if it is given).  For a block cursor, there should be
a maximum number of pixels that is affected.  If this number is
exceeded, a box cursor would be used with a thickness that does not
cause the number of pixels to exceed the threshold, but lets it have
at least the thickness of a box cursor + 1 pixel.

Something like that.  Yes, sounds tiresome.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

David Kastrup <dak@gnu.org> writes:

> Richard Stallman <rms@gnu.org> writes:
>
>>     > What exactly is the workaround you recommend?
>>
>>     Use :mask ...
>>
>> Then how about changing the default value for :mask?
>
> There is no sensible default.  What is useful as a mask depends on the
> image in question.  Using heuristic-mask can mean that the whole image
> flashes, but it can also mean that just a single corner pixel blinks.
>
> Even when a mask is specified, using the whole mask as a cursor
> indication is not a good idea, since a large image with a background
> will then flash horribly.  The right fix in my opinion would be to use
> the border of the image (extending into the image itself if the border
> is too small, overwriting the image when no mask is given, but maybe
> heeding the mask if it is given).  For a block cursor, there should be
> a maximum number of pixels that is affected.  If this number is
> exceeded, a box cursor would be used with a thickness that does not
> cause the number of pixels to exceed the threshold, but lets it have
> at least the thickness of a box cursor + 1 pixel.
>
> Something like that.  Yes, sounds tiresome.

Indeed.  So...

can we please postpone resolving this until after the release.

There is currently a work-around (using :mask) in the FEW (if any)
cases where there is no border around an image that could be used for
the cursor.  And the other work-around is to specify a border.

This is not a new bug -- it's been like this since 21.1, and only
recently have we had just one user complaining about it.  And he
seems satisfied with the changes I made to use a block cursor
for small images.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Ken Manheimer]

On 9/6/06, Kim F. Storm <storm@cua.dk> wrote:
> David Kastrup <dak@gnu.org> writes:
>
> > Richard Stallman <rms@gnu.org> writes:
> >
> >>     > What exactly is the workaround you recommend?
> >>
> >>     Use :mask ...
> >>
> >> Then how about changing the default value for :mask?
> >
> > There is no sensible default.  What is useful as a mask depends on the
> > image in question.  Using heuristic-mask can mean that the whole image
> > flashes, but it can also mean that just a single corner pixel blinks.
> >
> > Even when a mask is specified, using the whole mask as a cursor
> > indication is not a good idea, since a large image with a background
> > will then flash horribly.  The right fix in my opinion would be to use
> > the border of the image (extending into the image itself if the border
> > is too small, overwriting the image when no mask is given, but maybe
> > heeding the mask if it is given).  For a block cursor, there should be
> > a maximum number of pixels that is affected.  If this number is
> > exceeded, a box cursor would be used with a thickness that does not
> > cause the number of pixels to exceed the threshold, but lets it have
> > at least the thickness of a box cursor + 1 pixel.
> >
> > Something like that.  Yes, sounds tiresome.
>
> Indeed.  So...
>
> can we please postpone resolving this until after the release.
>
> There is currently a work-around (using :mask) in the FEW (if any)
> cases where there is no border around an image that could be used for
> the cursor.  And the other work-around is to specify a border.
>
> This is not a new bug -- it's been like this since 21.1, and only
> recently have we had just one user complaining about it.  And he
> seems satisfied with the changes I made to use a block cursor
> for small images.

the one catch is that i had not connected the :mask image spec with my
problem.  perhaps i'm dense, but i wonder whether the issue of
background, cursor, and image mask deserves some greater prominence in
the image documentation, so those floundering as i was get more
guidance.  as it currently stands, i wouldn't have connected the fact
that the mask can expose the "background of the frame" with the thing
i was trying to achieve, which is showing the cursor in the image.  (i
may have given it a try, but not been clear enough about whether to
expect the cursor to show through that i would notice whether or not
it worked.)

so i expect that at least some documentation improvement is in order.

-- 
ken
ken.manheimer@gmail.com
http://myriadicity.net

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    There is no sensible default.  What is useful as a mask depends on the
    image in question.  Using heuristic-mask can mean that the whole image
    flashes, but it can also mean that just a single corner pixel blinks.

1. So what?  Another default can't be worse than the present one,
   and it would be better MOST of the time.

2. Maybe we need another possible :mask value which would specify
   drawing the cursor in the way that you suggested.

3. If #2 is too complex to do now, maybe we should make it an error to
   fail to specify :mask on an image on a character in the buffer.
   That is simple.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

>     There is no sensible default.  What is useful as a mask depends on the
>     image in question.  Using heuristic-mask can mean that the whole image
>     flashes, but it can also mean that just a single corner pixel blinks.
>
> 1. So what?  Another default can't be worse than the present one,
>    and it would be better MOST of the time.

The present setting results in almost all cases in a recognizable
cursor position (images with border 0 are possible only come Emacs
22.1).  The indication is not perfect in all cases, but vastly better
than a large flashing image.  Which would happen to be the result of
the new default in most cases.  I have experienced that: trust me,
this default _is_ worse, much much worse than the existing one.  The
existing one is quirky, your suggestion would be result in it becoming
horrible.

preview-latex had to go to quite disadvantageous techniques to avoid
large-scale flashing in Emacs 21.  You'd reintroduce it.

> 2. Maybe we need another possible :mask value which would specify
> drawing the cursor in the way that you suggested.

I think it would be a mistake to abuse :mask just in order to get the
cursor to display in a useful way.  In particular since it would mean
that customizing the cursor type would have no effect on images, and
all image managing software would have to implement their own idea of
a cursor.

> 3. If #2 is too complex to do now, maybe we should make it an error to
>    fail to specify :mask on an image on a character in the buffer.
>    That is simple.

It will also break wagonloads of existing code.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    The present setting results in almost all cases in a recognizable
    cursor position (images with border 0 are possible only come Emacs
    22.1).

Are you saying that the problem of failing to show the cursor
only applies when the border is 0, and that that is a new feature?

In that case, let's limit my recommendations to the case when the
border is 0.  We can give :mask a different default, or make the lack
of :mask an error, only when border is 0.

      I have experienced that: trust me,
    this default _is_ worse, much much worse than the existing one.

That makes no sense at all.  The current default (in the case with
border 0) is not to show the cursor at all.  (If I'm mistaken, please
explain.)  If we change it to heuristic-mask, then usually it would
show the cursor.  That is better than not showing the cursor.

If you want me to believe this is worse, you'll have to demonstrate it
with clear explanation, not emotional upset.

    preview-latex had to go to quite disadvantageous techniques to avoid
    large-scale flashing in Emacs 21.  You'd reintroduce it.

Would you please show the justification for that claim?x

    I think it would be a mistake to abuse :mask just in order to get the
    cursor to display in a useful way.

There is no "abuse" in changing the default.

      In particular since it would mean
    that customizing the cursor type would have no effect on images, 

Why do you think it would mean that?

    > 3. If #2 is too complex to do now, maybe we should make it an error to
    >    fail to specify :mask on an image on a character in the buffer.
    >    That is simple.

    It will also break wagonloads of existing code.

I doubt it.  You've said border 0 is a new feature, so there can't
be so much existing code that uses it.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

>     The present setting results in almost all cases in a recognizable
>     cursor position (images with border 0 are possible only come Emacs
>     22.1).
>
> Are you saying that the problem of failing to show the cursor
> only applies when the border is 0, and that that is a new feature?

Yes.

> In that case, let's limit my recommendations to the case when the
> border is 0.  We can give :mask a different default, or make the
> lack of :mask an error, only when border is 0.

It is my opinion that we should not meddle with the defaults of mask,
since mask may serve independent purposes: it is well possible that
the _image_ has a transparency layer and specifies a mask.  And when
this mask just happens to be opaque (not an uncommon occurence), we
are back to square one: masks can perfectly well be all opaque.

The proper solution would be to draw in the image itself.  If we
instead use a temporary fix instead, it should not be interlocked with
:mask since the final, proper solution should be independent of mask,
too.

>       I have experienced that: trust me, this default _is_ worse,
>     much much worse than the existing one.
>
> That makes no sense at all.  The current default (in the case with
> border 0) is not to show the cursor at all.  (If I'm mistaken,
> please explain.)  If we change it to heuristic-mask, then usually it
> would show the cursor.  That is better than not showing the cursor.

We had cases in the combined Emacs/preview-latex history where the
cursor was not displayed (we have not used :border yet).  The user
will usually remember how he arrived at the character and not lose
track of the cursor.  If he gets away from the screen, a quick cursor
movement forth and back will reestablish the position.

The alternative is that the graphic on which the cursor is positioned
will be flashing (since the graphics tend to be characters on
background, this means the lion share of the graphics).

Now we have put the customization option for a flashing cursor into
the top of the options menu because people get close to epileptic
reactions by a 10x20 square (at most) flashing at them.

If a graphic taking half the screen or more is flashing (easily an
area larger by a factor of 1000), even the most insensitive person
will get close to getting a heart attack.  It is _very_ _very_
disconcerting, much more so than the cursor just disappearing.  The
cursor disappearing feels like a quirk: you can play around it, make
do by moving the cursor back and forth to figure out where you are.
The whole screen flashing at you feels like an attack on your sanity.
It is impossible to concentrate or do anything sensible except move
the cursor elsewhere.

> If you want me to believe this is worse, you'll have to demonstrate
> it with clear explanation, not emotional upset.
>
>     preview-latex had to go to quite disadvantageous techniques to avoid
>     large-scale flashing in Emacs 21.  You'd reintroduce it.
>
> Would you please show the justification for that claim?

In preview-latex under Emacs 21, we generate images with a colored
border and let just this border become transparent.  This means that
the images are useless for anything else.  There is a mode to cut and
paste a text passage into mail messages, and those color-border images
under Emacs-21 need to be regenerated without a border then.

The cursor display in Emacs 22 (a box around the cursor) was decent
enough to just use that: no artificial play with transparent borders
was needed.  The corresponding definition looks like the following:

    ;;; Note that the following default introduces a border only when
    ;;; Emacs blinks politely when point is on an image (the tested
    ;;; unrelated function was introduced at about the time image blinking
    ;;; became tolerable).
    (defcustom preview-transparent-border (unless (fboundp 'posn-object-x-y) 1.5)
      "Width of transparent border for previews in pt.
    Setting this to a numeric value will add a border of
    `preview-transparent-color' around images, and will turn
    the heuristic-mask setting of images to default to 't since
    then the borders are correctly detected even in case of
    palette operations.  If the transparent color is something
    not present otherwise in the image, the cursor display
    will affect just this border.  A width of 0 is interpreted
    by PostScript as meaning a single pixel, other widths are
    interpreted as PostScript points (1/72 of 1in)"
      :group 'preview-appearance
      :type '(choice (const :value nil :tag "No border")
                     (number :value 1.5 :tag "Border width in pt")))

>     I think it would be a mistake to abuse :mask just in order to get the
>     cursor to display in a useful way.
>
> There is no "abuse" in changing the default.

But :mask, :border and :cursor are unrelated things.  Making their
defaults depend on one another does not see like a good idea.

>     In particular since it would mean
>     that customizing the cursor type would have no effect on images, 
>
> Why do you think it would mean that?

There are cursor types of block and bar.  How would you display block
and bar cursors by fiddling with the default of :mask?

>     > 3. If #2 is too complex to do now, maybe we should make it an
>     > error to fail to specify :mask on an image on a character in
>     > the buffer.  That is simple.
>
>     It will also break wagonloads of existing code.
>
> I doubt it.  You've said border 0 is a new feature, so there can't
> be so much existing code that uses it.

Admitted.  I just don't think that making this an error is the right
migration path: ultimately, the cursor should just be drawn in the
image area.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     The present setting results in almost all cases in a recognizable
>     cursor position (images with border 0 are possible only come Emacs
>     22.1).
>
> Are you saying that the problem of failing to show the cursor
> only applies when the border is 0, and that that is a new feature?
>
> In that case, let's limit my recommendations to the case when the
> border is 0.  We can give :mask a different default, or make the lack
> of :mask an error, only when border is 0.

I just installed a change to always use the hollow cursor (which draws
inside the image border) for non-transparent images (with no mask).

It was a one-line change, not requiring changes to any platform files
(contrary to what I said earlier, we do already have the necessary
code to draw into the image -- namely the "draw hollow cursor" code).

I hope this settles the current issue.

--
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    I just installed a change to always use the hollow cursor (which draws
    inside the image border) for non-transparent images (with no mask).

"Inside the image border" has two natural meanings:

* Over the region that is the border

* Over a region that is smaller than the border and which the border
circumscribes.

The former would be a no-op with border 0; the latter would not.
Which one did you implement?

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

Richard Stallman <rms@gnu.org> writes:

>     I just installed a change to always use the hollow cursor (which draws
>     inside the image border) for non-transparent images (with no mask).
>
> "Inside the image border" has two natural meanings:
>
> * Over the region that is the border
>
> * Over a region that is smaller than the border and which the border
> circumscribes.
>
> The former would be a no-op with border 0; the latter would not.
> Which one did you implement?

It is a 1-pixel hollow rectangle along the "outer edge" of the area
occupied by the "image glyph" (whether it has a border, relief, both
or none) -- just like the hollow cursor on any other glyph.

So if the image has a border, the hollow cursor is in the border --
otherwise it is in the image itself.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Robert J. Chassell]

   It is a 1-pixel hollow rectangle along the "outer edge" ...

In other words, is it a 3 by 3 pixel square without color of any sort
in the central pixel, i.e., black in the central pixel, but some other
color, such as red in the other eight, of the nine pixels that make up
the square?

I am not sure what you mean by `1-pixel hollow'.  What do you mean
when you say `a 1-pixel hollow rectangle'?  Do you mean without color,
or do you mean transparent, that is to say, permitting the color of
what you refer to as ` the "outer edge" ' to show through, or
partially transparent (and if so, how much)?

Or do you mean that it is a one pixel square without color, that is to
say black, in that one pixel?  Or something else?  I do not grasp what
you mean.

Please be more precise for people like us, who see many alternatives.

-- 
    Robert J. Chassell                         
    bob@rattlesnake.com                         GnuPG Key ID: 004B4AC8
    http://www.rattlesnake.com                  http://www.teak.cc

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[tomas]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, Sep 10, 2006 at 11:00:49PM +0000, Robert J. Chassell wrote:
>    It is a 1-pixel hollow rectangle along the "outer edge" ...
> 
> In other words, is it a 3 by 3 pixel square without color of any sort
> in the central pixel [...]

I think I got it:

take tha whole image glyph (possibly with border, if it has one).

cut from it all around a strip, 1 pixel wide along the (rectangular)
edge.

This is the "area" of the cursor. It is a hollow rectangle whose
exterior is m x n pixels and whose interior is (m-1) x (n-1) pixels, if
the original glyph was m x n pixels.

HTH
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFFBPTlBcgs9XrR2kYRAot5AJ4hW5F7Ez7rblCaMQVgOvRotdcf5wCcDmSA
3dGDBM+OP/52p0zlCVsor8s=
=6+Wm
-----END PGP SIGNATURE-----

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Robert J. Chassell" <bob@rattlesnake.com> writes:

> Please be more precise for people like us, who see many alternatives.

Is this less ambiguous?

The cursor is shown as a 1-pixel wide line drawn along edge
of the combined image + border area.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[David Kastrup]

storm@cua.dk (Kim F. Storm) writes:

> "Robert J. Chassell" <bob@rattlesnake.com> writes:
>
>> Please be more precise for people like us, who see many alternatives.
>
> Is this less ambiguous?
>
> The cursor is shown as a 1-pixel wide line drawn along edge
> of the combined image + border area.

"[...] drawn inside the edge of [...]".  The edge itself is 0 pixels
wide and just between pixels.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Robert J. Chassell]

   The cursor is shown as a 1-pixel wide line drawn along edge
   of the combined image + border area.

Ah!  Am I right in understanding that you mean the cursor is a loop as
wide [first meaning of `wide'] as the combined image + border, and
that the loop is one pixel wide [second meaning of `wide']?

In a picture

    +----------------------+
    |                      |
    |     rest of          |
    |      image           |
    |                      |
    |                      |
    |                      |
    +----------------------+

    <---------------------->
      width, first meaning

The + - and | consist of the cursor, a loop.  

Am I right in understanding that you mean the width, second meaning,
of the loop is one pixel, or on one of my screens, a little less than
0.2 mm but in others more than 0.25 mm?

When you say the 1-pixel wide line is `hollow', what do you mean?

Do you mean the center of the loop is the image, i.e., the cursor does
not cover it up, and that the 1-pixel wide line is some color such as
black or red?  Or do you mean that pixels in the loop itself are
without color, meaning dark (trying to be black) on an RGB projection
system?  Or what?

-- 
    Robert J. Chassell                         
    bob@rattlesnake.com                         GnuPG Key ID: 004B4AC8
    http://www.rattlesnake.com                  http://www.teak.cc

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Kim F. Storm]

"Robert J. Chassell" <bob@rattlesnake.com> writes:

>    The cursor is shown as a 1-pixel wide line drawn along edge
>    of the combined image + border area.
>
> Ah!  Am I right in understanding that you mean the cursor is a loop as
> wide [first meaning of `wide'] as the combined image + border, and
> that the loop is one pixel wide [second meaning of `wide']?

Yes.

> Am I right in understanding that you mean the width, second meaning,
> of the loop is one pixel, or on one of my screens, a little less than
> 0.2 mm but in others more than 0.25 mm?

Yes, a pixel is a pixel is a pixel...

> When you say the 1-pixel wide line is `hollow', what do you mean?

It is similar to the other types of "hollow cursors" we have -- i.e.
the rest of the underlying glyph is still visible.

>
> Do you mean the center of the loop is the image, i.e., the cursor does
> not cover it up, 

Yes.

>                  and that the 1-pixel wide line is some color such as
> black or red?  

Yes, it drawn in the current "cursor color".

Have you never seen Emacs draw a hollow cursor before?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Richard Stallman]

    So if the image has a border, the hollow cursor is in the border --
    otherwise it is in the image itself.

Sounds good.  Thanks.

Re: cursor doesn't show through transparent images in emacs 22, unlike emacs 21

[Robert J. Chassell]

    > When you say the 1-pixel wide line is `hollow', what do you
    > mean?

    It is similar to the other types of "hollow cursors" we have --
    i.e.  the rest of the underlying glyph is still visible.

    > Do you mean the center of the loop is the image, i.e., the
    > cursor does not cover it up,

    Yes.

Now I understand.  Some of my pictures are big, and I did not realize
that you meant the cursor to go around them and that movement be
within a cursor.  The Emacs' hollow cursors I have seen before have
all been small, about one character in size.

    > ... and that the 1-pixel wide line is some color
    > such as black or red?

    Yes, it drawn in the current "cursor color".

My current "cursor color" is white, which may be a color in a picture,
so it could be not be seen.  Now you make sense.

Thank you.

-- 
    Robert J. Chassell                         
    bob@rattlesnake.com                         GnuPG Key ID: 004B4AC8
    http://www.rattlesnake.com                  http://www.teak.cc