font selection confusion

font selection confusion

[Miles Bader]

GNU Emacs 23.0.91.8 (x86_64-unknown-linux-gnu, GTK+ Version 2.14.7) of 2009-03-04 on dhlpc061

I'm kind of confused how emacs chooses alternative fonts to display
funny characters.

For instance, try this:

  (1) Start "HOME=/tmp emacs -Q"

  (2) Insert this string in *scratch*:  ❦⋮
      (those two characters are \u2766, "FLORAL HEART", and \u22ee,
      "VERTICAL ELLIPSIS")

  (3) On my system, which is debian unstable, the default font used is
      "Dejuavu Sans Mono"; using that font, the "FLORAL HEART" can be
      displayed using the default font, and "VERTICAL ELLIPSIS" is
      displayed using a different font,
      "-mutt-clearlyu-medium-r-normal--17-120-100-100-p-123-iso10646-1"

  (4) Now, I use the "Options > Set Default Font" menu to change the
      default font to "Liberation Mono" (this is a commonly installed
      free font, but any font with poor coverage of funny characters
      will seems to give the same result...).

  (5) Liberation Mono does not have either of those characters, but what
      happens is:  the "FLORAL HEART" character is now displayed as an
      empty box, but the "VERTICAL ELLIPSIS" character continues to be
      displayed using the "mutt-clearlyu" font.

Shouldn't emacs continue to use Dejavu Sans Mono to display "FLORAL
HEART", given that the font clearly contains it, and emacs clearly knows
(well, knew) that fact?
  
Thanks,

-Miles

-- 
=====
(^o^;
(()))
*This is the cute octopus virus, please copy it into your sig so it can spread.

Re: font selection confusion

[Stephen Berman]

On Wed, 04 Mar 2009 14:19:38 +0900 Miles Bader <miles.bader@necel.com> wrote:

> GNU Emacs 23.0.91.8 (x86_64-unknown-linux-gnu, GTK+ Version 2.14.7) of 2009-03-04 on dhlpc061
>
> I'm kind of confused how emacs chooses alternative fonts to display
> funny characters.
>
> For instance, try this:
>
>   (1) Start "HOME=/tmp emacs -Q"
>
>   (2) Insert this string in *scratch*:  ❦⋮
>       (those two characters are \u2766, "FLORAL HEART", and \u22ee,
>       "VERTICAL ELLIPSIS")
>
>   (3) On my system, which is debian unstable, the default font used is
>       "Dejuavu Sans Mono"; using that font, the "FLORAL HEART" can be
>       displayed using the default font, and "VERTICAL ELLIPSIS" is
>       displayed using a different font,
>       "-mutt-clearlyu-medium-r-normal--17-120-100-100-p-123-iso10646-1"

There's something peculiar with \u22ee in my Emacs (GNU Emacs 23.0.91.2
(i686-pc-linux-gnu, GTK+ Version 2.14.4) of 2009-03-01 on escher), which
may be related.  I also use Dejavu Sans Mono (set in ~/.Xresources,
along with xft as FontBackend).  In the Gnus *Article* buffer in which
I'm reading Miles's post, both characters are displayed fine.  According
to `C-u C-x =', both use the default font; here is the complete
description of \u22ee:

        character: ⋮ (8942, #o21356, #x22ee)
preferred charset: unicode-bmp (Unicode Basic Multilingual Plane (U+0000..U+FFFF))
       code point: 0x22EE
           syntax: . 	which means: punctuation
         category: .:Base
      buffer code: #xE2 #x8B #xAE
        file code: #xE2 #x8B #xAE (encoded by coding system utf-8-unix)
          display: by this font (glyph code)
    xft:-unknown-DejaVu Sans-normal-normal-normal-*-12-*-*-*-*-0-iso10646-1 (#xD13)

Character code properties: customize what to show
  name: VERTICAL ELLIPSIS
  general-category: Sm (Symbol, Math)

However, in this description, the vertical ellipsis is not displayed,
but instead an empty box.  The same goes for the Message buffer into
which I yanked the post from the *Article* buffer (again, the latter
does display the vertical ellipsis).  Likewise if I just yank that
character into any other buffer, and also if I start emacs -Q and insert
it with ucs-insert or with `M-: (insert ?\u22ee)': it displays only as
an empty box (again, with the default font Dejavu Sans Mono).  When I
type `C-u C-x =' on any of these empty boxes, this is the result:

        character: ⋮ (8942, #o21356, #x22ee)
preferred charset: unicode-bmp (Unicode Basic Multilingual Plane (U+0000..U+FFFF))
       code point: 0x22EE
           syntax: . 	which means: punctuation
         category: .:Base
      buffer code: #xE2 #x8B #xAE
        file code: #xE2 #x8B #xAE (encoded by coding system utf-8-unix)
          display: no font available

Character code properties: customize what to show
  name: VERTICAL ELLIPSIS
  general-category: Sm (Symbol, Math)


--
Steve Berman

Re: font selection confusion

[Kenichi Handa]

In article <buo63ip985x.fsf@dhlpc061.dev.necel.com>, Miles Bader <miles.bader@necel.com> writes:

>   (1) Start "HOME=/tmp emacs -Q"

>   (2) Insert this string in *scratch*:  ❦⋮
>       (those two characters are \u2766, "FLORAL HEART", and \u22ee,
>       "VERTICAL ELLIPSIS")

>   (3) On my system, which is debian unstable, the default font used is
>       "Dejuavu Sans Mono"; using that font, the "FLORAL HEART" can be
>       displayed using the default font, and "VERTICAL ELLIPSIS" is
>       displayed using a different font,
>       "-mutt-clearlyu-medium-r-normal--17-120-100-100-p-123-iso10646-1"

>   (4) Now, I use the "Options > Set Default Font" menu to change the
>       default font to "Liberation Mono" (this is a commonly installed
>       free font, but any font with poor coverage of funny characters
>       will seems to give the same result...).

>   (5) Liberation Mono does not have either of those characters, but what
>       happens is:  the "FLORAL HEART" character is now displayed as an
>       empty box, but the "VERTICAL ELLIPSIS" character continues to be
>       displayed using the "mutt-clearlyu" font.

This is because Emacs avoid greedy search of a font for each
entry of a fontset.  As those characters are both `symbol'
script, Emacs tries to find a font that has one of #x201C,
#x2200, or #x2500.  If a font is found, that font is used
for all symbol characters.  But some of found font may not
have a specific symbol character (e.g U+2766).  In such a
case, that character is shown by an empty box.

This is to avoid extreme slowness on displaying a buffer
that contains many characters for which you don't have any
font.

But, I've just installed a code that does a little bit more
greedy search.  It should works for such a font backend that
supports `has_char' method for a font-entity, e.g. xft.

---
Kenichi Handa
handa@m17n.org

Re: font selection confusion

[Miles Bader]

Kenichi Handa <handa@m17n.org> writes:
> But, I've just installed a code that does a little bit more
> greedy search.  It should works for such a font backend that
> supports `has_char' method for a font-entity, e.g. xft.

Ah, thank you!  I can see that it works by reading your message (in
gnus)!  :-)

-Miles

-- 
Friendless, adj. Having no favors to bestow. Destitute of fortune. Addicted to
utterance of truth and common sense.

Re: font selection confusion

[Kenichi Handa]

Sorry for the late response on this matter.

In article <87wsb5of23.fsf@escher.local.home>, Stephen Berman <stephen.berman@gmx.net> writes:

> There's something peculiar with \u22ee in my Emacs (GNU Emacs 23.0.91.2
> (i686-pc-linux-gnu, GTK+ Version 2.14.4) of 2009-03-01 on escher), which
> may be related.  I also use Dejavu Sans Mono (set in ~/.Xresources,
> along with xft as FontBackend).  In the Gnus *Article* buffer in which
> I'm reading Miles's post, both characters are displayed fine.  According
> to `C-u C-x =', both use the default font; here is the complete
> description of \u22ee:
[...]
> However, in this description, the vertical ellipsis is not displayed,
> but instead an empty box.  The same goes for the Message buffer into
> which I yanked the post from the *Article* buffer (again, the latter
> does display the vertical ellipsis).  Likewise if I just yank that
> character into any other buffer, and also if I start emacs -Q and insert
> it with ucs-insert or with `M-: (insert ?\u22ee)': it displays only as
> an empty box (again, with the default font Dejavu Sans Mono).  When I
> type `C-u C-x =' on any of these empty boxes, this is the result:

Do you still see this problem?  If so, please try this:

% emacs -Q
ESC : (setq font-log nil) RET
ESC : (insert ?\u22ee) RET
M-x font-show-log RET

and show me the contents of *Help* buffer.

---
Kenichi Handa
handa@m17n.org

Re: font selection confusion

[Stephen Berman]

On Thu, 19 Mar 2009 20:37:35 +0900 Kenichi Handa <handa@m17n.org> wrote:

> Sorry for the late response on this matter.
>
> In article <87wsb5of23.fsf@escher.local.home>, Stephen Berman <stephen.berman@gmx.net> writes:
>
>> There's something peculiar with \u22ee in my Emacs (GNU Emacs 23.0.91.2
>> (i686-pc-linux-gnu, GTK+ Version 2.14.4) of 2009-03-01 on escher), which
>> may be related.  I also use Dejavu Sans Mono (set in ~/.Xresources,
>> along with xft as FontBackend).  In the Gnus *Article* buffer in which
>> I'm reading Miles's post, both characters are displayed fine.  According
>> to `C-u C-x =', both use the default font; here is the complete
>> description of \u22ee:
> [...]
>> However, in this description, the vertical ellipsis is not displayed,
>> but instead an empty box.  The same goes for the Message buffer into
>> which I yanked the post from the *Article* buffer (again, the latter
>> does display the vertical ellipsis).  Likewise if I just yank that
>> character into any other buffer, and also if I start emacs -Q and insert
>> it with ucs-insert or with `M-: (insert ?\u22ee)': it displays only as
>> an empty box (again, with the default font Dejavu Sans Mono).  When I
>> type `C-u C-x =' on any of these empty boxes, this is the result:
>
> Do you still see this problem?  If so, please try this:

No, in my latest build (GNU Emacs 23.0.91.4 (i686-pc-linux-gnu, GTK+
Version 2.14.4) of 2009-03-15 on escher) the character displays fine in
all buffers.  So I guess the problem has been fixed.

Steve Berman

Re: font selection confusion

[Kenichi Handa]

In article <87k56lrayo.fsf@escher.local.home>, Stephen Berman <stephen.berman@gmx.net> writes:

>>> There's something peculiar with \u22ee in my Emacs (GNU Emacs 23.0.91.2
>>> (i686-pc-linux-gnu, GTK+ Version 2.14.4) of 2009-03-01 on escher), which
>>> may be related.  I also use Dejavu Sans Mono (set in ~/.Xresources,
>>> along with xft as FontBackend).  In the Gnus *Article* buffer in which
>>> I'm reading Miles's post, both characters are displayed fine.  According
>>> to `C-u C-x =', both use the default font; here is the complete
>>> description of \u22ee:
> > [...]
>>> However, in this description, the vertical ellipsis is not displayed,
>>> but instead an empty box.  The same goes for the Message buffer into
>>> which I yanked the post from the *Article* buffer (again, the latter
>>> does display the vertical ellipsis).  Likewise if I just yank that
>>> character into any other buffer, and also if I start emacs -Q and insert
>>> it with ucs-insert or with `M-: (insert ?\u22ee)': it displays only as
>>> an empty box (again, with the default font Dejavu Sans Mono).  When I
>>> type `C-u C-x =' on any of these empty boxes, this is the result:
> >
> > Do you still see this problem?  If so, please try this:

> No, in my latest build (GNU Emacs 23.0.91.4 (i686-pc-linux-gnu, GTK+
> Version 2.14.4) of 2009-03-15 on escher) the character displays fine in
> all buffers.  So I guess the problem has been fixed.

I see.  Thank you for confiming that.

---
Kenichi Handa
handa@m17n.org