And the identical JDK 7 backport ..
http://cr.openjdk.java.net/~prr/8015556.jdk7/
-phil.
On 5/31/13 3:40 AM, Andrew Brygin wrote:
The change looks fine, and I have not noticed any artifacts.
Thanks,
Andrew
On 5/31/2013 12:53 AM, Phil Race wrote:
http://cr.openjdk.java.net/~prr/8015556/
http://bugs.sun.com/view_bug.do?bug_id=8015556
Supplementary chars are DOA in the OS X port since
the mapper code there is ignorant of them, doing such
things as down-casting an int code point to a char before
doing the lookup and storing the first char of a surrogate pair
as being the map key, not the decoded supplementary char.
Interestingly the key to the map was of type integer so
that part I didn't have to change.
I've run through Font2DTest and SwingSet with this patch.
Since it touches a key part of the code if there's bugs I'd
expect them to show up there. I didn't see any problems
and it improves the supplementary char support.
I expect to back port this to 7u40 since its clearly a major issue,
but there's not much time for it to bake in 8 first, so please
try it out (if you can).
-phil.