--===============1272627663== Content-Type: multipart/alternative; boundary="0000000000008da85105bc092742" --0000000000008da85105bc092742 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Tue, Feb 23, 2021 at 1:00 AM Werner LEMBERG wrote: > > > Can you give an example? We were talking about examples like combining a base glyph with a rare mark. If the mark is missing in the font, you have different options what to do - look for a precomposed glyph, find a suitable variant of the mark, or go for all fallback font. Maybe Behdad has concrete examples. Again, please give an example. It seems to me that you are actually > thinking of improving Pango, not fontconfig per se. I only wonder > whether the approach you are envisioning has drawbacks to applications > that are not using Pango. > Sure, this is about improving font rendering. The high-level parts of it ar= e implemented in Pango. But it needs suitable font enumeration apis, which is where fontconfig comes in. I don't object to using harfbuzz =E2=80=93 this is what FreeType is actuall= y > using for getting glyph coverage for its auto-hinter. However, > harfbuzz has some quite massive dependencies that might not be wanted > by the user. > Great, this might be a more or less solved problem , then. From the Pango perspective, a harfbuzz dependency is not such a big deal - we use harfbuzz anyway. And using harfbuzz directly, instead of via freetype, might have other advantages for fontconfig. Matthias --0000000000008da85105bc092742 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable


=
On Tue, Feb 23, 2021 at 1:00 AM Werne= r LEMBERG <wl@gnu.org> wrote:


Can you give an example?

We were talking ab= out examples like combining a base glyph with a
rare mark. If the= mark is missing in the font, you have different options
what to = do - look for a precomposed glyph, find a suitable variant of the
mark, or go for all fallback font. Maybe Behdad has concrete examples.
=

Again, please give an example.=C2=A0 It seems to me that you are actually thinking of improving Pango, not fontconfig per se.=C2=A0 I only wonder
whether the approach you are envisioning has drawbacks to applications
that are not using Pango.

Sure, this is= about improving font rendering. The high-level parts of it are
i= mplemented in Pango. But it needs suitable font enumeration apis, which
is where fontconfig comes in.

I don't object to using harfbuzz =E2=80=93 this is what FreeType is act= ually
using for getting glyph coverage for its auto-hinter.=C2=A0 However,
harfbuzz has some quite massive dependencies that might not be wanted
by the user.

Great, this might be a mor= e or less solved problem , then.

From the Pango pe= rspective, a harfbuzz dependency is not such a big
deal - we use = harfbuzz anyway. And using harfbuzz directly, instead of
via free= type, might have other advantages for fontconfig.

= Matthias
--0000000000008da85105bc092742-- --===============1272627663== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Fontconfig mailing list Fontconfig@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/fontconfig --===============1272627663==--