Lost antialias?

Lost antialias?

[Angelo Graziosi]

The following happens on GNU/Linux Kubuntu 8.04.

Curent trunk (a few minutes ago) looks to have lost the antialias. Now
Emacs appears horrible. The build I did less than 24 hours ago looks 
very fine.

Cheers,
    Angelo.


PS: I have tried to attach the figures, but the emails bounced. Sorry.

Re: Lost antialias?

[Kenichi Handa]

In article <485230EB.9070203@alice.it>, Angelo Graziosi <angelo.graziosi@alice.it> writes:

> The following happens on GNU/Linux Kubuntu 8.04.
> Curent trunk (a few minutes ago) looks to have lost the antialias. Now
> Emacs appears horrible. The build I did less than 24 hours ago looks 
> very fine.

Perhaps it's because of my change, but by just "horrible", I
don't know what's going on.  At least, in my environment,
xft backend works well with antialias.

Please start Emacs as this:

% EMACS_FONT_LOG=1 emacs -Q

and show me the *Help* buffer made by M-x font-show-log RET.

And, also show me the result of C-u C-x = on a normal ASCII
character.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Angelo Graziosi]

Kenichi Handa ha scritto:
> In article <485230EB.9070203@alice.it>, Angelo Graziosi <angelo.graziosi@alice.it> writes:
> 
>> The following happens on GNU/Linux Kubuntu 8.04.
>> Curent trunk (a few minutes ago) looks to have lost the antialias. Now
>> Emacs appears horrible. The build I did less than 24 hours ago looks 
>> very fine.
> 
> Perhaps it's because of my change, but by just "horrible", I
> don't know what's going on.  At least, in my environment,
> xft backend works well with antialias.
> 
> Please start Emacs as this:
> 
> % EMACS_FONT_LOG=1 emacs -Q
> 
> and show me the *Help* buffer made by M-x font-show-log RET.

Sorry but 'font-show-log' seems do not exist! only font-lock...

> 
> And, also show me the result of C-u C-x = on a normal ASCII
> character.
> 

C-u C-x = a  prints 'a' etc..


It seems that there is some problem with 'DejaVu Sans Mono-12' font and 
friends (BitStream Vera Sans Mono).

Usually I start Emacs:

$ emacs23 --font "DejaVu Sans Mono-12" &

and this gives the 'horrible' (i.e. the characters appear not smoothed) 
look.

Instead if I use "Monospace-12"

$ emacs23 --font "Monospace-12" &

the result is identical to that obtained until 24 hours ago with 
'DejaVu...', i.e. very fine.

I did the build:

${source_dir}/configure --prefix=${prefix_dir}
make bootstrap


Cheers,
    Angelo.


PS: May someone explain why when I attach PNG images the email bounces 
from emacs-devel? I did this some month ago and it worked.



> ---
> Kenichi Handa
> handa@ni.aist.go.jp

Re: Lost antialias?

[Thomas Baumann]

I'm stepping in:

xfont-list: -*-DejaVu Sans Mono-*-*-*-*-*-*-*-*-*-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--*-0-0-m-*-iso8859-1
ftfont-list: -*-DejaVu Sans Mono-*-iso8859-1
  -unknown-DejaVu Sans Mono-bold-oblique-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-normal-oblique-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-bold-normal-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-normal-normal-normal-*-m-0-iso8859-1
list: -*-DejaVu Sans Mono-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--*-0-0-m-*-iso8859-1
  -unknown-DejaVu Sans Mono-bold-oblique-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-normal-oblique-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-bold-normal-normal-*-m-0-iso8859-1
  -unknown-DejaVu Sans Mono-normal-normal-normal-*-m-0-iso8859-1
sort-by: -*-DejaVu Sans Mono-normal-normal-normal-*-12-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
open: -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--12-87-100-100-m-70-iso8859-1
xfont-list: -misc-dejavu sans mono-*-*-*--*-*-*-*-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-bold-i-normal--*-0-0-m-*-iso8859-1
ftfont-list: -misc-dejavu sans mono-*--*-m-*-iso8859-1
list: -misc-dejavu sans mono-medium-*-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
sort-by: -misc-dejavu sans mono-medium-italic-normal--12-*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
open: -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--12-87-100-100-m-70-iso8859-1
list: -misc-dejavu sans mono-*-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
sort-by: -misc-dejavu sans mono-light-r-normal--12-*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
open: -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--12-87-100-100-m-70-iso8859-1
list: -misc-dejavu sans mono-*-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
sort-by: -misc-dejavu sans mono-light-r-normal--12-*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
list: -misc-dejavu sans mono-*-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
sort-by: -misc-dejavu sans mono-bold-r-normal--12-*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
open: -misc-dejavu sans mono-bold-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-bold-r-normal--12-87-100-100-m-70-iso8859-1
list: -misc-dejavu sans mono-medium-*-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-o-normal--17-*-100-100-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-r-normal--17-*-100-100-m-*-iso8859-1
sort-by: -misc-dejavu sans mono-medium-italic-normal--12-*-0-0-m-*-iso8859-1
  -misc-dejavu sans mono-medium-i-normal--*-0-0-m-*-iso8859-1


and C-u C-x = on a s:


        character: s (115, #o163, #x73)
preferred charset: ascii (ASCII (ISO646 IRV))
       code point: 0x73
           syntax: w 	which means: word
         category: a:ASCII graphic characters 32-126 (ISO646 IRV:1983[4/0]) l:Latin r:Japanese roman
      buffer code: #x73
        file code: not encodable by coding system iso-latin-9-unix
          display: by this font (glyph code)
     -misc-dejavu sans mono-medium-r-normal--12-87-100-100-m-70-iso8859-1 (#x73)

Character code properties are not shown: customize what to show

There are text properties here:
  auto-composed        t

Thomas

RE: Lost antialias?

[Christian.Lynbech]

I can confirm this behaviour. I have just updated and recompiled and
suddenly my fonts were no longer antialiased.

It appears that renaming 'mono' to 'monospace' as suggested fixes the
problem. Ie. if one uses "DejaVu Sans Monospace-12" rather than "DejaVu
Sans Mono-12", antialiasing comes back. 

Possibly something in font name parsing has changed. 

			-- Christian Lynbech

Re: Lost antialias?

[Angelo Graziosi]

Kenichi Handa ha scritto:
> In article <485230EB.9070203@alice.it>, Angelo Graziosi <angelo.graziosi@alice.it> writes:
> 
>> The following happens on GNU/Linux Kubuntu 8.04.
>> Curent trunk (a few minutes ago) looks to have lost the antialias. Now
>> Emacs appears horrible. The build I did less than 24 hours ago looks 
>> very fine.
> 
> Perhaps it's because of my change, but by just "horrible", I
> don't know what's going on.  At least, in my environment,
> xft backend works well with antialias.
> 

I have tried also on Cygwin[1], and the result is different.

On Cygwin I have /.Xdefauls:

Emacs.FontBackend:      xft
Emacs.font:             Bitstream Vera Sans Mono-14

Xft*antialias: true
Xft*autohint: true

The look is fine (better?), but the height is 13 not 14.

Now, to obtain the height 14 as with previous build, I must put

Emacs.font:             Bitstream Vera Sans Mono-15

in /.Xdefaults.


On GNU/Linux Kubuntu 8.04 the result is still 'horrible' with "DejaVu 
Sans Mono-12" and "BitStream Vera Sans Mono", but is fine with 
"Monospace-12".

On Kubuntu I do not use /.Xdefaults but I start Emacs as

emacs23 --font "DejaVu Sans Mono-12" &


Cheers,
    Angelo.


---
[1] ${source_dir}/configure --prefix=${prefix_dir} \
                             --with-x-toolkit=lucid

Re: Lost antialias?

[Kenichi Handa]

In article <48546807.8010406@alice.it>, Angelo Graziosi <angelo.graziosi@alice.it> writes:

> I have tried also on Cygwin[1], and the result is different.

> On Cygwin I have /.Xdefauls:

> Emacs.FontBackend:      xft
> Emacs.font:             Bitstream Vera Sans Mono-14

> Xft*antialias: true
> Xft*autohint: true

> The look is fine (better?), but the height is 13 not 14.

The font size problem is because of the bug introduced by
gtk-style fontname parser as I wrote in the previous mail.

And I think "fine" means you have antialise font, then
perhaps it's because you use xft backend only.

> On GNU/Linux Kubuntu 8.04 the result is still 'horrible' with "DejaVu 
> Sans Mono-12" and "BitStream Vera Sans Mono", but is fine with 
> "Monospace-12".

> On Kubuntu I do not use /.Xdefaults but I start Emacs as

> emacs23 --font "DejaVu Sans Mono-12" &

I see.  So, you have "dejavu sans mono" font both by X and
Xft backend.  And, as I recently changed the font selector
to prefer a font supplied by a backend of higher priority
(the default order is x,xft on X Window System), you get
"dejavu sans mono" via X backend.  When you specify
"monospace" instead, X backend doesn't know about it, but
xft backend knows that it's a generic family name, and your
system assignes "dejavu sans mono" to it.  Thus you get
antialias font via xft backend.

If you never want "dejavu sans mono" font as a X font, just
remove it from your font-path.

Or, set X resource as 'Emacs*FontBackend: xft, x'.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Chong Yidong]

Kenichi Handa <handa@m17n.org> writes:

> The font size problem is because of the bug introduced by
> gtk-style fontname parser as I wrote in the previous mail.

Yes.  I believe this should now work properly.

>> On GNU/Linux Kubuntu 8.04 the result is still 'horrible' with "DejaVu 
>> Sans Mono-12" and "BitStream Vera Sans Mono", but is fine with 
>> "Monospace-12".
>
>> On Kubuntu I do not use /.Xdefaults but I start Emacs as
>
>> emacs23 --font "DejaVu Sans Mono-12" &
>
> I see.  So, you have "dejavu sans mono" font both by X and
> Xft backend.  And, as I recently changed the font selector
> to prefer a font supplied by a backend of higher priority
> (the default order is x,xft on X Window System), you get
> "dejavu sans mono" via X backend.  When you specify
> "monospace" instead, X backend doesn't know about it, but
> xft backend knows that it's a generic family name, and your
> system assignes "dejavu sans mono" to it.  Thus you get
> antialias font via xft backend.

Is there any reason to have x,xft by default, instead of xft,x?

Re: Lost antialias?

[Kenichi Handa]

In article <87abhmgmko.fsf@stupidchicken.com>, Chong Yidong <cyd@stupidchicken.com> writes:

> Is there any reason to have x,xft by default, instead of xft,x?

Backward compatibility.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Miles Bader]

Kenichi Handa <handa@m17n.org> writes:
>> Is there any reason to have x,xft by default, instead of xft,x?
>
> Backward compatibility.

Is this really very important?

I suspect many (maybe most) people will _want_ the "incompatible" font
when there's a conflict.

-Miles

-- 
It wasn't the Exxon Valdez captain's driving that caused the Alaskan oil spill.
It was yours.  [Greenpeace advertisement, New York Times, 25 February 1990]

Re: Lost antialias?

[Kenichi Handa]

In article <buove09riwd.fsf@dhapc248.dev.necel.com>, Miles Bader <miles.bader@necel.com> writes:

> Kenichi Handa <handa@m17n.org> writes:
>>> Is there any reason to have x,xft by default, instead of xft,x?
> >
> > Backward compatibility.

> Is this really very important?

> I suspect many (maybe most) people will _want_ the "incompatible" font
> when there's a conflict.

Until the code gets more stable, I'd like to keep backward
compatibility where it's easy to do that.

And, at least, I think Xft backend is slower when you runs
Emacs remotely via narrow network.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Stefan Monnier]

>>>> Is there any reason to have x,xft by default, instead of xft,x?
>> > Backward compatibility.
>> Is this really very important?
>> I suspect many (maybe most) people will _want_ the "incompatible" font
>> when there's a conflict.
> Until the code gets more stable, I'd like to keep backward
> compatibility where it's easy to do that.

What's the recommended way to change this option in .emacs?


        Stefan

Re: Lost antialias?

[Kenichi Handa]

In article <jwvy755qr2w.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>>>> Is there any reason to have x,xft by default, instead of xft,x?
>>> > Backward compatibility.
>>> Is this really very important?
>>> I suspect many (maybe most) people will _want_ the "incompatible" font
>>> when there's a conflict.
> > Until the code gets more stable, I'd like to keep backward
> > compatibility where it's easy to do that.

> What's the recommended way to change this option in .emacs?

(push '(font-backend xft x) default-frame-alist)

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Chong Yidong]

Kenichi Handa <handa@m17n.org> writes:

> In article <87abhmgmko.fsf@stupidchicken.com>, Chong Yidong <cyd@stupidchicken.com> writes:
>
>> Is there any reason to have x,xft by default, instead of xft,x?
>
> Backward compatibility.

Will re-ordering the font backends fix the other problem we discussed,
which is that the `fixed-pitch' and `variable-pitch' faces are
non-fontconfig fonts?  We need to solve this problem, because it looks
bad if Emacs acquires Fontconfig font support but the startup screen
doesn't use those fonts.

Re-ordering the font-backends in .emacs does not solve the problem.

Should we simply abolish the use of `fixed-pitch' and `variable-pitch'?

Re: Lost antialias?

[Miles Bader]

Chong Yidong <cyd@stupidchicken.com> writes:
> Should we simply abolish the use of `fixed-pitch' and `variable-pitch'?

What do you propose to replace `variable-pitch'?

-Miles

-- 
o The existentialist, not having a pillow, goes everywhere with the book by
  Sullivan, _I am going to spit on your graves_.

Re: Lost antialias?

[Stefan Monnier]

>>> Is there any reason to have x,xft by default, instead of xft,x?
>> Backward compatibility.

> Will re-ordering the font backends fix the other problem we discussed,
> which is that the `fixed-pitch' and `variable-pitch' faces are
> non-fontconfig fonts?  We need to solve this problem, because it looks
> bad if Emacs acquires Fontconfig font support but the startup screen
> doesn't use those fonts.

> Re-ordering the font-backends in .emacs does not solve the problem.

> Should we simply abolish the use of `fixed-pitch' and `variable-pitch'?

I think this would not be an appropriate solution.  It's more akin to
hiding the problem than to solving it.  Even more so since users's
.emacs as well as external package may make use of those faces.


        Stefan

Re: Lost antialias?

[Angelo Graziosi]

I would remember that since this problem appeared on GNU/Linux (June 
12/13) there is a related problem on Cygwin [1].

In my .Xdefaults:

------------------------------------------------------------------
Emacs.FontBackend:      xft
Emacs.font:             Bitstream Vera Sans Mono-14

URxvt*font:             xft:Bitstream Vera Sans
                         Mono:size=14:medium:roman:antialias=true

Xft*antialias: true
Xft*autohint: true
------------------------------------------------------------------

Before June 12/13, in both Emacs and Urxvt the font had the same height 
14. Now the font in Emacs has a shorter height, 13, while in Urxvt is 
still 14. In other word in Emacs is all more little, now. On Cygwin the 
build of Emacs uses XFT and Lucid (on Cygwin GTK does not works).

Now I should use Bitstream Vera Sans Mono-15 to have the same result.


Cheers,
    Angelo.

---
[1] http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg01068.html
     http://lists.gnu.org/archive/html/emacs-devel/2008-06/msg01127.html

Re: Lost antialias?

[Chong Yidong]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>>> Is there any reason to have x,xft by default, instead of xft,x?
>>> Backward compatibility.
>
>> Will re-ordering the font backends fix the other problem we discussed,
>> which is that the `fixed-pitch' and `variable-pitch' faces are
>> non-fontconfig fonts?  We need to solve this problem, because it looks
>> bad if Emacs acquires Fontconfig font support but the startup screen
>> doesn't use those fonts.
>
>> Re-ordering the font-backends in .emacs does not solve the problem.
>
>> Should we simply abolish the use of `fixed-pitch' and `variable-pitch'?
>
> I think this would not be an appropriate solution.  It's more akin to
> hiding the problem than to solving it.  Even more so since users's
> .emacs as well as external package may make use of those faces.

How about treating "Helv" and "Courier" specially in
ffont_resolve_generic_family, similar to "Sans" and "Mono"?

Re: Lost antialias?

[Miles Bader]

Chong Yidong <cyd@stupidchicken.com> writes:
> How about treating "Helv" and "Courier" specially in
> ffont_resolve_generic_family, similar to "Sans" and "Mono"?

Why don't we just change the default face definitions to something more
appropriate on a modern system?

Maybe it would be also useful to support something like web-browsers
use, where you could specify a "family" of "Dejavu Sans,Helvetica,Sans",
and it would split the list at the commas and use the first one it could
find.

That way we could safely use "Dejavu Sans" [etc] when available, which
is probably the nicest and most consistent widely available choice on
modern FOSS systems.

-Miles

-- 
Carefully crafted initial estimates reward you not only with
reduced computational effort, but also with understanding and
increased self-esteem.         -- Numerical methods in C,
  Chapter 9. "Root Finding and Nonlinear Sets of Equations"

Re: Lost antialias?

[Chong Yidong]

Miles Bader <miles@gnu.org> writes:

> Chong Yidong <cyd@stupidchicken.com> writes:
>> How about treating "Helv" and "Courier" specially in
>> ffont_resolve_generic_family, similar to "Sans" and "Mono"?
>
> Why don't we just change the default face definitions to something more
> appropriate on a modern system?

That could work too.  How about combining the defface redefinition with
a change to face-font-family-alternatives so that this will work on
Windows and Mac?  (I'm not sure what the appropriate fonts are for those
platforms, though.)

(defface fixed-pitch
  '((t :family "Monospace"))
  "The basic fixed-pitch face."
  :group 'basic-faces)

(defcustom face-font-family-alternatives
  '(("Monospace" "DejaVu Sans Mono" [WINDOWS AND MAC FONTS]  "fixed")
    ....))

What font should we use for variable-pitch?  Bitstream Vera Sans?

> Maybe it would be also useful to support something like web-browsers
> use, where you could specify a "family" of "Dejavu Sans,Helvetica,Sans",
> and it would split the list at the commas and use the first one it could
> find.

That could be useful, if Handa (or someone else) wants to implement it.
Otherwise, I guess it's not crucial.

Re: Lost antialias?

[Miles Bader]

Chong Yidong <cyd@stupidchicken.com> writes:
> What font should we use for variable-pitch?  Bitstream Vera Sans?

"Dejavu Sans" is probably the best if it's available.  It's the same as
Bitstream Vera Sans (more or less a copy AIUI) except that it has
greater coverage.

-Mile

-- 
Mayonnaise, n. One of the sauces that serve the French in place of a state
religion.

Re: Lost antialias?

[Stefan Monnier]

> (defface fixed-pitch
>   '((t :family "Monospace"))
>   "The basic fixed-pitch face."
>   :group 'basic-faces)

> (defcustom face-font-family-alternatives
>   '(("Monospace" "DejaVu Sans Mono" [WINDOWS AND MAC FONTS]  "fixed")
>     ....))

> What font should we use for variable-pitch?  Bitstream Vera Sans?

I'd put "Sans-Serif" first, so as to re-use fontconfig's default.
Similarly for fixed-pitch (not sure if that's "mono" or "monospace" or
what).

>> Maybe it would be also useful to support something like web-browsers
>> use, where you could specify a "family" of "Dejavu Sans,Helvetica,Sans",
>> and it would split the list at the commas and use the first one it could
>> find.

We can already do exactly that with an additional indirection through
face-font-family-alternatives, and indeed that's exactly what we do for
fixed-pitch and variable-pitch.


        Stefan

Re: Lost antialias?

[Miles Bader]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>> What font should we use for variable-pitch?  Bitstream Vera Sans?
>
> I'd put "Sans-Serif" first, so as to re-use fontconfig's default.
> Similarly for fixed-pitch (not sure if that's "mono" or "monospace" or
> what).

Fontconfig's generic handling seems aimed more at finding something
"usable" rather than something "good", and the ordering seems almost
random (Bitstream Vera comes before Dejavu, even though the latter is
more or less a better version of the former, and on my system "Sans
Serif" resolves to "Kochi Gothic", apparently simply because I have that
installed).

So I think using generic names as the first choice would often give
suboptimal results.

-Miles

-- 
Generous, adj. Originally this word meant noble by birth and was rightly
applied to a great multitude of persons. It now means noble by nature and is
taking a bit of a rest.

Re: Lost antialias?

[Stefan Monnier]

>>> What font should we use for variable-pitch?  Bitstream Vera Sans?
>> I'd put "Sans-Serif" first, so as to re-use fontconfig's default.
>> Similarly for fixed-pitch (not sure if that's "mono" or "monospace" or
>> what).

> Fontconfig's generic handling seems aimed more at finding something
> "usable" rather than something "good", and the ordering seems almost
> random (Bitstream Vera comes before Dejavu, even though the latter is
> more or less a better version of the former, and on my system "Sans
> Serif" resolves to "Kochi Gothic", apparently simply because I have that
> installed).

Hmm... so what do other apps use?  I think we should aim to use whatever
other applications use by default, since there's a better chance that
this has been correctly setup (and if not, at least we won't be worse
than those other apps and we can just say "sorry, not our fault").

E.g. what does Firefox use?  (or Vim ;-)?


        Stefan

Re: Lost antialias?

[Miles Bader]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>>> What font should we use for variable-pitch?  Bitstream Vera Sans?
>>> I'd put "Sans-Serif" first, so as to re-use fontconfig's default.
>>> Similarly for fixed-pitch (not sure if that's "mono" or "monospace" or
>>> what).
>
>> Fontconfig's generic handling seems aimed more at finding something
>> "usable" rather than something "good", and the ordering seems almost
>> random (Bitstream Vera comes before Dejavu, even though the latter is
>> more or less a better version of the former, and on my system "Sans
>> Serif" resolves to "Kochi Gothic", apparently simply because I have that
>> installed).
>
> Hmm... so what do other apps use?  I think we should aim to use whatever
> other applications use by default, since there's a better chance that
> this has been correctly setup (and if not, at least we won't be worse
> than those other apps and we can just say "sorry, not our fault").
>
> E.g. what does Firefox use?  (or Vim ;-)?

Actually I was mistaken about "Kochi Gothic" -- it seems to use it
because I have LANG set to "ja_JP".

If I use:

   LANG=C HOME=/tmp fc-match "Serif"

then I get "Bitstream Vera Serif" (etc for "Sans", and "Mono").

So I guess the choice is "reasonable", except for the ordering of Vera
before Dejavu.  I don't know the reason for that offhand, but I imagine
perhaps it's because Vera is used as a general default these days, and
Dejavu is something still installed as an optional package.

Using "Mono", "Sans", or "Serif" as families all return similar results,
so I suppose they would be reasonable defaults.

Most of the substitution seems to be done by config files in /etc/fonts,
and I'm not sure how things are on other systems; I'm using Debian
unstable.

[Whether the above changing of default for LANG=ja_JP is really
desirable or not, I dunno, but I guess it's going to yield more familiar
results when mixing CJK with ascii characters.  Anyway, as you say,
"not our problem".]

-Miles

-- 
Mayonnaise, n. One of the sauces that serve the French in place of a state
religion.

Re: Lost antialias?

[Eli Zaretskii]

> From: Miles Bader <miles@gnu.org>
> Date: Sat, 21 Jun 2008 13:52:10 +0900
> Cc: Chong Yidong <cyd@stupidchicken.com>, angelo.graziosi@alice.it,
> 	Kenichi Handa <handa@m17n.org>, emacs-devel@gnu.org
> 
> Whether the above changing of default for LANG=ja_JP is really
> desirable or not, I dunno, but I guess it's going to yield more familiar
> results when mixing CJK with ascii characters.

I actually think that it might be not a good idea to change the
default font selection depending on the LANG setting.  What are the
advantages of such an approach?  Are there other GUI applications that
behave similarly?

Re: Lost antialias?

[Miles Bader]

Eli Zaretskii <eliz@gnu.org> writes:
>> Whether the above changing of default for LANG=ja_JP is really
>> desirable or not, I dunno, but I guess it's going to yield more familiar
>> results when mixing CJK with ascii characters.
>
> I actually think that it might be not a good idea to change the
> default font selection depending on the LANG setting.  What are the
> advantages of such an approach?  Are there other GUI applications that
> behave similarly?

It's fontconfig that does it, not Emacs.

-Miles

-- 
Opposition, n. In politics the party that prevents the Goverment from running
amok by hamstringing it.

Re: Lost antialias?

[Eli Zaretskii]

> From: Miles Bader <miles@gnu.org>
> Date: Sat, 21 Jun 2008 15:56:36 +0900
> Cc: cyd@stupidchicken.com, emacs-devel@gnu.org, monnier@iro.umontreal.ca,
> 	angelo.graziosi@alice.it, handa@m17n.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> >> Whether the above changing of default for LANG=ja_JP is really
> >> desirable or not, I dunno, but I guess it's going to yield more familiar
> >> results when mixing CJK with ascii characters.
> >
> > I actually think that it might be not a good idea to change the
> > default font selection depending on the LANG setting.  What are the
> > advantages of such an approach?  Are there other GUI applications that
> > behave similarly?
> 
> It's fontconfig that does it, not Emacs.

But Emacs calls it, doesn't it?  So we could set up things for the
results to be consistent across LANG settings, if we decide that to be
TRT.

The question remains: what is TRT?

Re: Lost antialias?

[Miles Bader]

Eli Zaretskii <eliz@gnu.org> writes:
>> It's fontconfig that does it, not Emacs.
>
> But Emacs calls it, doesn't it?  So we could set up things for the
> results to be consistent across LANG settings, if we decide that to be
> TRT.

I don't necessarily disagree with that, but the question is whether such
an effort to "be consistent" would incur too much cost -- for instance,
inconsistency with other modern apps, or ignoring user-specified global
settings.

I think that Stefan has a good point when he says we should rely on
existing mechanisms to the extent that they work acceptably.

I'd like to know what e.g., Gnome apps do in this case, but I'm not sure
how to get Gnome to ignore my settings and use its defaults (even if I
change HOME, still seems to use my settings).

-Miles

-- 
Religion, n. A daughter of Hope and Fear, explaining to Ignorance the nature
of the Unknowable.

Re: Lost antialias?

[Eli Zaretskii]

> From: Miles Bader <miles@gnu.org>
> Cc: handa@m17n.org,  emacs-devel@gnu.org
> Date: Sat, 21 Jun 2008 17:29:26 +0900
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> >> It's fontconfig that does it, not Emacs.
> >
> > But Emacs calls it, doesn't it?  So we could set up things for the
> > results to be consistent across LANG settings, if we decide that to be
> > TRT.
> 
> I don't necessarily disagree with that, but the question is whether such
> an effort to "be consistent" would incur too much cost -- for instance,
> inconsistency with other modern apps, or ignoring user-specified global
> settings.
> 
> I think that Stefan has a good point when he says we should rely on
> existing mechanisms to the extent that they work acceptably.
> 
> I'd like to know what e.g., Gnome apps do in this case

100% agreement.  That's precisely why I asked in my original mail how
do other GUI applications behave wrt LANG settings.

Re: Lost antialias?

[Stephen J. Turnbull]

Eli Zaretskii writes:

 > > > I actually think that it might be not a good idea to change the
 > > > default font selection depending on the LANG setting.  What are the
 > > > advantages of such an approach?  Are there other GUI applications that
 > > > behave similarly?
 > > 
 > > It's fontconfig that does it, not Emacs.
 > 
 > But Emacs calls it, doesn't it?  So we could set up things for the
 > results to be consistent across LANG settings, if we decide that to be
 > TRT.

You don't want to do that.  The immediate counterexample is Japanese
and Chinese.  There is a complete overlap in repertoire for the
Japanese syllabary, as well as a 90% overlap in the most common hanzi.
This means that there will be no way currently available to Emacs to
detect the "right" font for text/plain;charset=utf-8 unless you pay
attention to locale.  This matters; Japanese and Chinese have quite
different tastes in glyphs, and as a non-reader of Chinese and a
non-native reader of Japanese, I have to say that to me, Chinese fonts
look more like pointillist art than text.

From the XEmacs sources:

    /* #### It's my branch, my favorite charsets get checked first!
       That's a joke, Son.
       Ie, I don't know what I'm doing, so my charsets first is as good as
       any other arbitrary order.  If you have a better idea, speak up! */

I've been ragged on about this on several occasions over the last
three or four years, but I still don't have a better idea---and
neither do my detractors.

I don't claim to have worked on it systematically, but with that
history do you think it's worth the hard work it will take to come up
with a "good" generic ordering?  That's up to you, of course.

Re: Lost antialias?

[James Cloos]

>>>>> "Miles" == Miles Bader <miles@gnu.org> writes:

Miles> So I guess the choice is "reasonable", except for the ordering of
Miles> Vera before Dejavu.  I don't know the reason for that offhand,
Miles> but I imagine perhaps it's because Vera is used as a general
Miles> default these days, and Dejavu is something still installed as an
Miles> optional package.

Keith wrote about that on the fontconfig list.

IIRC, the issue was that the quality of the glyphs added by DejaVu might
not be the same as those in Vera, and perhaps other fonts should take
precedence for languages and scripts in that catagory.

i tend to disagree, but my impression was that he got feedback from
distributions—who had gotten feedback from their userbase—essentially
requesting the current choice.

Personally, I leave Vera out of the aliases on my box and rely on just
DejaVu for those outlines.

(Written in variable-pitch-mode using DejaVu Serif. :)

(Although, interestingly, it does not show up in the list of minor modes
in the mode line.  Only (Message SC MML Abbref Fill Narrow) are there.
Is there a limit on how many are included in the string?)

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Lost antialias?

[Miles Bader]

James Cloos <cloos@jhcloos.com> writes:
> IIRC, the issue was that the quality of the glyphs added by DejaVu might
> not be the same as those in Vera, and perhaps other fonts should take
> precedence for languages and scripts in that catagory.
>
> i tend to disagree, but my impression was that he got feedback from
> distributions—who had gotten feedback from their userbase—essentially
> requesting the current choice.

What's funny is that the few times I've noticed the difference --
usually an oddly accented character in the middle of a name -- the
Vera+Something-Else-Random rendering was egregiously, stupendously,
awful (e.g., a badly-scaled fixed-with font with serifs, in the middle
of a Vera Sans string, at twice the height...).

I'm betting on "Our users request that nothing at all be changed, ever"
myself...

-miles

-- 
We live, as we dream -- alone....

Re: Lost antialias?

[Werner LEMBERG]

> IIRC, the issue was that the quality of the glyphs added by DejaVu
> might not be the same as those in Vera, and perhaps other fonts
> should take precedence for languages and scripts in that catagory.

Indeed.  The newly added glyphs lack proper TrueType bytecode hinting.
If you use the autohinter of FreeType, there shouldn't be a
visible difference.


    Werner

Re: Lost antialias?

[Kenichi Handa]

In article <874p7n1jb9.fsf@catnip.gol.com>, Miles Bader <miles@gnu.org> writes:

> [Whether the above changing of default for LANG=ja_JP is really
> desirable or not, I dunno, but I guess it's going to yield more familiar
> results when mixing CJK with ascii characters.  Anyway, as you say,
> "not our problem".]

We can add "LANG=en" in FcPattern to get the effect of
LANG=en in ftfont.c.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Kenichi Handa]

In article <87lk0z1w3e.fsf@catnip.gol.com>, Miles Bader <miles@gnu.org> writes:

> Why don't we just change the default face definitions to something more
> appropriate on a modern system?

I think that is good.  For instance, when we change
variable-pitch to family "sans", fontconfig surely finds a
proper font, and by adding ("sans" "helvetica" "helv" ...)
to face-font-family-alternatives, X can also find some font.
And for fixed-pitch, "mono" is good.

> Maybe it would be also useful to support something like web-browsers
> use, where you could specify a "family" of "Dejavu Sans,Helvetica,Sans",
> and it would split the list at the commas and use the first one it could
> find.

This is a good idea in general, but for the problem of
"variable-pitch" and "fixed-pitch", I think using "sans" and
"mono" are better.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Lost antialias?

[Stefan Monnier]

>> Maybe it would be also useful to support something like web-browsers
>> use, where you could specify a "family" of "Dejavu Sans,Helvetica,Sans",
>> and it would split the list at the commas and use the first one it could
>> find.

> This is a good idea in general, but for the problem of
> "variable-pitch" and "fixed-pitch", I think using "sans" and
> "mono" are better.

As mentioned, you don't have to say "but" because using
face-font-family-alternatives gets us *exactly* the above.
So, could someone change the fixed-pitch and variable-pitch faces and
face-font-family-alternatives accordingly to use (as first choice) the
most generic corresponding fontconfig name.

BTW, I'm not sure what are the official generic fontconfig names: my
/etc/fonts config files seems to say that "sans" and "mono" are
deprecated and should be replaced by "sans-serif" and "monospace".


        Stefan


PS: Curiously, it seems that the serif/sans distinction is not taken
very seriously for monospace fonts, since I don't see any "mono-sans" or
"mono-serif" in fontconfig, and indeed our own face-font-family-alternatives
puts together "courier" and "fixed".

Re: Lost antialias?

[Chong Yidong]

I've changed faces.el so that variable-pitch uses "Sans" and fixed-pitch
uses "Mono".  Fontconfig should automatically find a family, either
DejaVu or something else.  For systems without fontconfig, I added
aliases for "Sans" and "Mono" in face-font-family-alternatives.

Re: Lost antialias?

[James Cloos]

BTW, these comments are from the default fonts.conf:

,----
| <!--
|   Accept deprecated 'mono' alias, replacing it with 'monospace'
| -->
| <!--
|   Accept alternate 'sans serif' spelling, replacing it with 'sans-serif'
| -->
| <!--
|   Accept deprecated 'sans' alias, replacing it with 'sans-serif'
| -->
`----

So the names probably should be "monospace" and "sans-serif".

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Lost antialias?

[Chong Yidong]

James Cloos <cloos@jhcloos.com> writes:

> BTW, these comments are from the default fonts.conf:
>
> ,----
> | <!--
> |   Accept deprecated 'mono' alias, replacing it with 'monospace'
> | -->
> | <!--
> |   Accept alternate 'sans serif' spelling, replacing it with 'sans-serif'
> | -->
> | <!--
> |   Accept deprecated 'sans' alias, replacing it with 'sans-serif'
> | -->
> `----
>
> So the names probably should be "monospace" and "sans-serif".

Okay, I changed it.  Thanks.

Re: Lost antialias?

[Angelo Graziosi]

Kenichi Handa ha scritto:
> In article <48546807.8010406@alice.it>, Angelo Graziosi <angelo.graziosi@alice.it> writes:
> 
>> I have tried also on Cygwin[1], and the result is different.
> 
>> On Cygwin I have /.Xdefauls:
> 
>> Emacs.FontBackend:      xft
>> Emacs.font:             Bitstream Vera Sans Mono-14
> 
>> Xft*antialias: true
>> Xft*autohint: true
> 
>> The look is fine (better?), but the height is 13 not 14.
> 
> The font size problem is because of the bug introduced by
> gtk-style fontname parser as I wrote in the previous mail.
> 
> And I think "fine" means you have antialise font, then
> perhaps it's because you use xft backend only.
> 
>> On GNU/Linux Kubuntu 8.04 the result is still 'horrible' with "DejaVu 
>> Sans Mono-12" and "BitStream Vera Sans Mono", but is fine with 
>> "Monospace-12".
> 
>> On Kubuntu I do not use /.Xdefaults but I start Emacs as
> 
>> emacs23 --font "DejaVu Sans Mono-12" &
> 
> I see.  So, you have "dejavu sans mono" font both by X and
> Xft backend.  And, as I recently changed the font selector
> to prefer a font supplied by a backend of higher priority
> (the default order is x,xft on X Window System), you get
> "dejavu sans mono" via X backend.  When you specify
> "monospace" instead, X backend doesn't know about it, but
> xft backend knows that it's a generic family name, and your
> system assignes "dejavu sans mono" to it.  Thus you get
> antialias font via xft backend.
> 
> If you never want "dejavu sans mono" font as a X font, just
> remove it from your font-path.
> 
> Or, set X resource as 'Emacs*FontBackend: xft, x'.
> 

OK, I will use, on GNU/Linux, Monospace which gives the same result.

But on Cygwin, a fresh build (a few minutes ago), still gives height 13 
using 'Bitstream Vera Sans Mono-14'? or is this value, I used for 
months, wrong and now I should set it to 15 to have the same result?

Cheers,
    Angelo.