[ft] Re: [ft-devel] FreeType 2.3.0 release candidate 1 is available

[Mike FABIAN]

"David Turner" <address@hidden> さんは書きました:

> Hello everyone,
>
> just to inform you that FreeType 2.3.0rc1 is now available at
> the following web addresses:

Thank you, this fixes also

    http://bugzilla.novell.com/show_bug.cgi?id=231417

Packages of FreeType 2.3.0rc1 for openSUSE 10.2 are available
at

    http://software.opensuse.org/download/M17N/openSUSE_10.2/

(I named the packages freetype2-2.2.1.20070112 to get monotonously
increasing version numbers)

> ================copy of docs/CHANGES============================
> CHANGES BETWEEN 2.3.0 and 2.2.1
>
>   I. IMPORTANT BUG FIXES

[...]

>     - Fixed the TrueType bytecode  loader to deal properly with subtle
>       monochrome/gray  issues  when   scaling  the  CVT.   Some  fonts
>       exhibited bad rendering artifacts otherwise.

I don't know whether this is related, but I found that the
rendering of the font "Consolas" from Windows Vista has become worse.

A screen shot "rendering-regression.png" is attached.

On the left side is the rendering of "Consolas" with freetype 2.3.0rc1
and on the right side with freetype 2.2.1.20061027 (CVS snapshot from
20061027).

Both screen shots are made with full hinting and with using the
byte code interpreter.  In both cases FIX_BYTECODE is commented
out at the top of ttinterp.c (done by a patch in freetype
2.2.1.20061027, freetype 2.3.0rc1 already comes like that).

The older version of freetype renders "Consolas" much sharper.  The
characters 'M', 'V', 'W', don't look good when rendered by the old
version either, but overall the old result looks much better to me.

>     - The unpatented  hinter is now part  of the default  build of the
>       library; we  have added  code to automatically  support `tricky'
>       fonts that need it.

>       This means  that FreeType should `just work'  with certain Asian
>       fonts, like  MingLiU, which cannot properly be  loaded without a
>       bytecode interpreter,  but which fortunately  do not use  any of
>       the patented  bytecode opcodes.  We detect these  fonts by
>       name,
        ^^^^^

I have 2 versions of MingLiU here, the older one needs this
hack, the newer one doesn't and works OK even without the byte
code interpreter.

The 2 versions are:

    (c) Copyright DynaLab Inc. 1992-2001MingLiUMingLiURegularVersion
    3.21Trademark by DynaLab Inc.

    (c) Copyright DynaComware Corp. 2005MingLiURegularVersion
    6.01Trademark by DynaComware Corp.

The name is the identical.

But probably there is no other possibility than to detect these fonts
by name.

-- 
Mike FABIAN   <address@hidden>   http://www.suse.de/~mfabian
睡眠不足はいい仕事の敵だ。
I � Unicode

rendering-regression.png
Description: PNG image