Hi,
> Can you set my expectations
ok ..
There won't be any remedy in 7u40. Its 99% baked.
Beyond that I don't know and can't make promises since work would have to be
done and I don't have any release schedule - even though I would love
to improve this.
Also remember this is a forum for people who are *contributing* fixes &
changes,
not a support forum.
If you have a fix that might work and doesn't compromise security,
please propose
it - assuming you have signed the Oracle Contributor Agreement.
-phil.
On 7/26/13 2:52 PM, Keith Kriewall wrote:
Hi Phil,
Can you set my expectations for if/when this problem will be fixed? We can't
upgrade past 7u21 until this gets corrected, or we find some kind of
workaround. Right now I can't envision a workaround for a Web Start
environment.
Thanks,
-Keith
-----Original Message-----
From: Phil Race [mailto:philip.r...@oracle.com]
Sent: Tuesday, July 02, 2013 10:47 AM
To: Keith Kriewall
Cc: '2d-dev@openjdk.java.net'
Subject: Re: [OpenJDK 2D-Dev] 7u25 kills 'blit' performance for any custom
ColorModel class
This is an unfortunate consequence of a fix a potential security issue.
The actual bug ID for this is 8019590. We'll take a look to see what can be
done.
-phil.
On 7/2/2013 10:13 AM, Keith Kriewall wrote:
Hi 2d Devs,
The addition of BufferedImage.isStandard() has absolutely killed
drawing performance for our application. BufferedImage types that used
to be set to optimized types (e.g. TYPE_INT_BGR) are now set to
TYPE_CUSTOM, due solely to the classloader used to load the class. It
appears that unless we load our subclasses of ColorModel via the
bootclassloader, we're stuck. I've posted 9004714 to Oracle's bug
database with a test app.
Can anyone on this list comment on this change in 7u25, provide a
workaround, and/or provide a fix to bug 9004714 for future 7u releases?
Thanks,
-Keith
*Keith Kriewall*
Software Developer
705 5^th Ave S, Suite 1100
Seattle, WA 98104 USA
206-217-7892
keith.kriew...@attachmate.com <mailto:keith.kriew...@attachmate.com>
Description: line
Description: Attachmate <http://www.attachmate.com/>
