[PATCH 0/4] Fix and improve Freetype/Fontconfig support

[Andrew Hughes]

----- Original Message -----
> Hello @all,
> 
> as suggested by Andrew Haley (thanks!), here my little patch series for
> fixing and improving the Freetype/Fontconfig support.
> 
> All patches are against icedtea8-forest/jdk.
> 

I'll open a bug for getting these into 3.3.0.

> Once those patches are applied (or even just the first one), it is my
> suggestion to rename the Fontconfig support entirely (configure option
> and its associated C preprocessor symbol) from infinality to fontconfig
> or something alike as well as enable it by default.

Can you elaborate on what kind of option you would expect the new version
to use and what tests would be required to ensure it will build and run
correctly against the installed FreeType?

Currently, we detect if TT_CONFIG_OPTION_SUBPIXEL_HINTING is set to anything
other than zero. This can mean infinality support is installed, or just
the new stripped-down 2.7 version.

> 
> The reasoning behind it: The fixed freetypeScaler contains nothing that
> would require a patched Freetype nor Fontconfig. The crashes and/or
> rendering anomalies seen were caused by an uninitialized ftRenderMode
> which is fixed with patch one in the series. I tested both Freetype 2.7
> and specially 2.6.3 (prior to the bci revamp of 2.6.4) w/o infinality
> enabled. No problems at all.
> Also, since this generally improves the user experience and should have
> no ill side-effects, there is no harm in enabling it by default.
> Last but not least, the current naming is thus misleading and also does
> not really convey what is behind it.
> 
> If there are any problems, suggestions or anything else, please let me
> know and I will gladly fix it.
> 

Are you open to maintaining this going forward? The existing code was
imported into IcedTea 2.x as-is from patches people were already applying
to the code, and it seemed better to have the code in there rather than
having users patching it in. All I added was the option to not build the
code. It's not something I have great interest or use for personally, but
I'm aware that others do.

Initially, it was enabled by default, but was then disabled by default after
users reported issues. I think it should have some time to soak before we
consider enabling it by default again.

> Thanks for taking the time,
> Matthias
> 
> --
> Dipl.-Inf. (FH) Matthias Dahl | Software Engineer | binary-island.eu
>  Hire me for contract work: Open Source, Proprietary, Web/Mobile/Desktop
> 
> 

-- 
Andrew :)

Senior Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

PGP Key: ed25519/0xCFDA0F9B35964222 (hkp://keys.gnupg.net)
Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222