Quick update on project status

[Roman Kennke]

Hi Igor,

> >> At the moment proposed fix for replacing encumbered T2K library with 
> >> freetype is under code review.
> >> Upon approval (it is over 3500 lines and review will likely take some 
> >> time and several iterations)
> >> changes will start making their way to master workspace and going public.
> >> Depending on integration windows schedule this will take few weeks to 
> >> land on openjdk.java.net.
> >>     
> >
> > Great! So this is will be a complete replacement for the encumbered T2K
> > library?
> >   
> Yes, there will be no t2k binary library in the OpenJDK and OepnJDK will 
> be runnable
> (at least it works for me :).

Very cool! I was wondering, how other font stuff is implemented in
(Open)JDK, most importantly TextLayout. Are you using some external
implementation for that (I think freetype doesn't cover TextLayout), or
an own implementation? In GNU Classpath we had an own implementation,
based on java.text.Bidi, but it might make sense to use some external
lib (like HarfBuzz).

> There could be compatibility issues (have yet to pass JCK) but these 
> should be possible to fix.
> We can not enable truetype hinting by default though.

Hmm. I think ideally there should be an option to dynamically link
freetype and do whatever the system's freetype installation does. I
think that this might be problematic wrt JCK, but maybe not. The
advantages of this are better resource usage (the library and fonts are
very likely already loaded, at least on Unix systems), less redundancy,
better security handling (only needs fixing the freetype lib if problems
arise and not the JDK installation too, and of course, if the system
freetype installation has truetype hinting enabled, the JDK would
benefit from this.

> >> Changes introduce abstraction of scaler engine (sun.font.FontScaler) and 
> >> implementations for Freetype and T2K.
> >> They were tested with freetype 2.3.4 but likely should work with all 
> >> 2.3.* versions.
> >>     
> >
> > 'Abstraction of scaler engine', does that mean that it would be possible
> > to provide alternative implementations? I would be very interested in
> > that (I told you in Brussels).
> >
> >   
> Theoretically - yes.
> So far there are 3 implementations - t2k font scaler, freetype font 
> scaler and null font scaler.
> Selection logic is hardcoded for now but this can be revised later and 
> easy to change in source code.

Very cool! As soon as I can look at the code, I'll see how 'my' 100%
Java font implementation can plug into this, and make proposals on how
to improve the abstraction (if necessary).

Cheers, Roman

-- 
http://kennke.org/blog/