On Sat, Jun 30, 2018 at 3:03 AM, Werner LEMBERG <[email protected]> wrote:

>
> >> > Third, "palette[layer_color_index]" is recipe for invalid memory
> >> > access.
> >>
> >> Is it?  The code checks that `layer_color_index' is not out of
> >> bound.  Do you have a better idea?
> >
> > What does the code do if the font has bad index then?
>
> It simply returns 0, as documented.
>
> > Maybe treating those as foreground works.
>
> Mhmm.  I could implement that, but is this a good idea?  Wouldn't it
> be better if the rendering engine drops invalid color layered glyphs,
> returning the `standard' outline instead?
>
> >> What I could do is to make the magic number 0xFFFF completely
> >> disappear by giving it index `num_palette_entries' while increasing
> >> the size of `palette' (and `num_palette_entries') by one element...
> >
> > I'd say stay close to the spec.  Just document returning 0xFFFF.
>
> OK.
>
> > Ok, bigger problem.  I'm getting crashes rendering anything using
> > hb-view: [...]
>
> Please repeat your test with
>
> * compiling with `-O0 -g3',
> * setting `FT2_DEBUG=any:7', and
>

This needs enabling debugging, right?


> * running `hb-view' with valgrind.
>

Running in gdb and valgrind both points to the same stacktrace I sent.
Inspecting in gdb, sfnt->get_colr_layer is NULL but called on ftobjs.c:5496


>
>   Werner
>



-- 
behdad
http://behdad.org/
_______________________________________________
Freetype-devel mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/freetype-devel

Reply via email to