Looks like you are right. Its 4 years since the relevant code went in to
JDK 7
but I had remembered the intent as being that distro+major version needed to
match on Linux. So on any RHEL 6 update RHEL.6.properties and
RHEL6.1.properties should
all result in a hit but not RHEL.properties. Maybe I decided instead
that the smoke test to verify
that at least some of the files were present was sufficient and useful
to help cases
where we didn't want to be forced to provide ones for each version.
Still, for JDK 7 and 8 perhaps you'd be better off to delete all the
files to escape the maintenance
After checking that the result of "fc-match -s sans" etc is close to
what you'd hard code anyway ..
BTW I don't think JCK can validly test this since there's no conformance
requirement
to have any particular international language support for display.
-phil.
On 5/25/2012 4:19 AM, Jiri Vanek wrote:
On 05/24/2012 06:14 PM, Phil Race wrote:
How did you verify this ? I ask because this file will be (or should
be!)
completely ignored by JDK 7 and 8, since it does not have a name that
precisely references Fedora.16 - ie distro AND variant.
I have testfile which generate few hundred Chinese/Indian/Korean/Japan
chars in bold/italic/normal and in different styles
(Serif/Sanserif/Dialog....). Without my patch a lot of characters were
not rendered correctly. With - only two of them have failed. Imho - it
would be nice to have it in jck.
AFIK there is fall back in fontocnfig - it tries
linux.fontconfig.Fedora.VERSION.properties first, if not found then
linux.fontconfig.Fedora.properties if nto found then
linux.fontconfig.roperties (if not found then
fontconfig.properties?). For icedtea6 I have patched
jdk/make/sun/awt/Makefile to have
linux.fontconfig.Fedora.properties as default with configuration for
most recent fedora and had copies with older Fedoras' fontconfigs. Yah
- a lot of maintenance! We had also patch to support subversions (eg
linux.fontconfig.Rhel.6.3.properties x
linux.fontconfig.Rhel.6.2.properties x
linux.fontconfig.Rhel.6.properties x linux.fontconfig.Rhel.properties)
Because I have not seen something similar in Openjdk7 i used this
default - linux.fontconfig.Fedora.properties and it worked as expected.
In general I am not sure there is even a point to maintaining this
file since
we have code in JDK 7 that pulls the logical font definitions from
/usr/lib/fontconfig
APIs at runtime and so we should match the native desktop without the
maintenance
burden.
This is awesome! Thanx so much for finally replacing java's
"artificial font configuration". When this change will reach release?
Maybe I can push my change to used branches only...
But for sure I will follow your advices.
The only reason I can think of for having these files any more is if
you consciously
want something very different than GTK/Qt/etc is using ..
I definitely don't think this is what I want:)
J.
-phil.
On 5/24/2012 12:45 AM, Jiri Vanek wrote:
Hi all!
http://jvanek.fedorapeople.org/oracle/jdk7/webrev/ and
http://jvanek.fedorapeople.org/oracle/jdk8/webrev/ (actually they
are same) are my fixes for
Fedora's fontconfig.
They are sync with newest stable release.
Although this is my first contribution to OpenJDK I have pushed this
change to IcedTea6 already
some time ago.
I have successfully tested this on OpenJDK7 head i386 and x86_64. I
was not able to verify jdk8
because I have failed to Build OpenJDK8 (But I'm working on it!),
but AFAIK the fontconfig
Still not done. Few time left and gcc in f17 behaving out of my scope :(.
algorithm have not changed.
Also I have noticed that there is no
linux.fontconfig.Rhel.properties. I have handled this one for
IcedTea6 too. How do you see possible addition of this file?
Looking forward to contribute to OpenJDK7/8
Best regards
Jiri Vanek
