Re: [ft-devel] Incremental font support

[Henry Stiles]

On Sun, Feb 22, 2009 at 4:49 AM, Werner LEMBERG <address@hidden> wrote:
>
>> Remove the 'if ( glyph_index >= face->root.num_glyphs )' clause at
>> t1gload.c:287, and the test example renders.
>>
>> So, what can we do to fix this? The check was added when the glyph
>> id bounds checking as moved from FT_Load_Glyph to resolve
>> http://savannah.nongnu.org/bugs/?18301.
>
> The main question: What does `glyph_index' stand for within the bridge
> to GhostScript?  In another recent mail Jun Dai asked whether
> FT_Load_Glyph can load real glyph indices for CID-keyed fonts, and
> I've answered `no'.  Probably you need exactly the same.
>
> If this my assumption is correct I'll add a load flag to FT_Load_Glyph
> to suppress the special CID handling.
>
>
>    Werner
>

I just returned from a business trip and must have missed this mail.
CID fonts are something I need to look at more closely before I can
answer completely.  See this open bug from Hirot Kagotani for
instance:

http://bugs.ghostscript.com/show_bug.cgi?id=689826

The final problem I found with freetype testing is charpaths were not
scaled correctly.  In postscript it is possible to return fonts as
outlines without rasterizing.  Our code is configured to expect
freetype to use a 16.16 fixed point representation which apparently it
does for some things, returned outlines use 26.6.  With that fix we
seem to be in pretty good shape.

I have to assume the community has only been using the freetype bridge
for CID fonts, without reverting the 2 changes in freetype and scaling
the charpaths correctly I don't see how anything would print
reasonably.

We'll try to follow up on the CID question this week, thanks for all your help.

Henry