Re: Add a test for bitmap font metrics to ensure that they match the Windows ones

[Huw Davies]

On Fri, Mar 17, 2006 at 02:01:41PM +0800, Dmitry Timoshkov wrote:
> "Huw D M Davies" <h.davies1(a)physics.ox.ac.uk> wrote:
> > Looks like this might be a fontforge or FreeType problem.  Windows
> > courier at 13 ppem should have an ascent of 11 and a descent of 2,
> > which is what we have in the .sfd.  Now, when I run sfnt2fnt on the
> > generated .ttf I get an ascent of 13, ppem of 13 and hence a descent
> > of 0.
> >
> > I'm not going to have time to look into this soon, so please feel free
> > to investigate more.
>
> Looks like it's a sfnt2fnt bug and not a fontforge or FreeType problem.
>
> This patch tests bitmap font metrics to ensure that they match the Windows
> ones. The patch discovers a bug in sfnt2fnt which causes Courier font to
> have wrong ascent and descent values (13/0 instead of 11/2). courier.ttf
> generated by fontforge from courier.sfd has correct values, but courier-*.fnt
> files have wrong values.

Well we get the ascent from the height of the Aring glyph.  As far as
I can tell this is 11 in fontforge.  So somewhere between fontforge's
export as a ttf and sfnt2fnt reading that ttf (using FreeType) the
value has changed to 13.  Hence my comment that either fontforge or
FreeType is doing something odd.

Huw.
-- 
Huw Davies
huw(a)codeweavers.com

Add a test for bitmap font metrics to ensure that they match the Windows ones

[Dmitry Timoshkov]

"Huw Davies" <huw(a)codeweavers.com> wrote:
> > Looks like it's a sfnt2fnt bug and not a fontforge or FreeType problem.
> >
> > This patch tests bitmap font metrics to ensure that they match the Windows
> > ones. The patch discovers a bug in sfnt2fnt which causes Courier font to
> > have wrong ascent and descent values (13/0 instead of 11/2). courier.ttf
> > generated by fontforge from courier.sfd has correct values, but courier-*.fnt
> > files have wrong values.
>
> Well we get the ascent from the height of the Aring glyph.  As far as
> I can tell this is 11 in fontforge.  So somewhere between fontforge's
> export as a ttf and sfnt2fnt reading that ttf (using FreeType) the
> value has changed to 13.  Hence my comment that either fontforge or
> FreeType is doing something odd.

I was going to point out that I tried to replace in tools/sfnt2fnt.c

ascent = face->glyph->metrics.height >> 6;

by

ascent = face->size->metrics.ascender >> 6;

which is what is used in dlls/gdi/freetype.c,get_bitmap_text_metrics(),
but that breaks MS Sans Serif bitmap fonts. It looks like fontforge
doesn't allow to have a different font header for each different bitmap
size in the bitmap font.

-- 
Dmitry.