Re: [1/4] gdi32: Fixup the internal ntmFlags for bitmap fonts.

[Huw Davies]

On Thu, Nov 12, 2015 at 10:07:56PM +0900, Akihiro Sagawa wrote:
> This fixes the test failure on font.c:389 with the next patch.
>
> The failure caused by the potential bug that Western System font is a
> fake bold.

Could you explain what this bug is?

Apart from that, while I think this series looks ok, I'm uncomfortable
with this going in before 1.8 (the fact that is uses new FreeType
functionality makes me especially nervous).  So I'm afraid you'll have
to hold off until after the release.

Sorry about that,

Huw.

[1/4] gdi32: Fixup the internal ntmFlags for bitmap fonts.

[Akihiro Sagawa]

On Mon, 16 Nov 2015 10:57:03 +0000, Huw Davies wrote:
> > The failure caused by the potential bug that Western System font is a
> > fake bold.
> > Could you explain what this bug is?

Sure. Currently, the fake bold flag is governed by ntmFlags. If ntmFlags
isn't set NTM_BOLD bit and the font weight, specified by a user, is
greater than 550 (see freetype.c, line 5555), fake bold flag is
activated (line 5601).
ntmFlags bits are set by get_ntm_flags function. If ft_face->style_flags
& FT_STYLE_FLAG_BOLD is true, then NTM_BOLD flag is set (line 1947).
For TrueType or OpenType fonts, the flag works as expected.  However,
this flag isn't set for bold Windows bitmap fonts (e.g. System; weight
600) in FreeType library unless weight >= 800[1].
Therefore, the fake_bold flag is incorrectly set for System bitmap font,
although the font is a real bold font.
[1] ... http://git.savannah.gnu.org/cgit/freetype/freetype2.git/tree/src/winfonts/wi...
> Apart from that, while I think this series looks ok, I'm uncomfortable
> with this going in before 1.8 (the fact that is uses new FreeType
> functionality makes me especially nervous).  So I'm afraid you'll have
> to hold off until after the release.
> I see. I'll resubmit the series of patches after 1.8 release.

Thanks for reviewing,
Akihiro Sagawa