Hi Martin, There was no need for creating separate test case to override BufferedImage and see what happens but its good that we know the behaviour now. I had already approved webrev.01 and still +1 for webrev.01.
And I think we should conclude the thread with Phil’s question about just returning 0 in these methods. Thanks, Jay > On 25-Jun-2020, at 8:33 PM, Martin Desruisseaux > <martin.desruisse...@geomatys.com> wrote: > > Hello > > Following up on JDK-8166038 [1], in case it may help to justify the proposed > fix, attached is another test case demonstrating the bug. Attempt to draw a > BufferedImage subimage with Graphics2D causes an exception to be thrown if we > hide the fact that the image is a BufferedImage instance, for preventing > SunGraphics2D to use its special cases for BufferedImage. Stack trace is: > > Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: > BufferedImages only have one tile with index 0,0 > at > java.desktop/java.awt.image.BufferedImage.getTile(BufferedImage.java:1401) > at SubimageDrawingTest$Wrapper.getTile(SubimageDrawingTest.java:136) > at > java.desktop/sun.java2d.SunGraphics2D.drawTranslatedRenderedImage(SunGraphics2D.java:2821) > at > java.desktop/sun.java2d.SunGraphics2D.drawRenderedImage(SunGraphics2D.java:2708) > at SubimageDrawingTest.main(SubimageDrawingTest.java:31) > Replacing getTileGridXOffset() and getTileGridYOffset() return values by 0 > (as expected and as requested by method contract) fix the bug. > > This bug happens with any code that try to handle RenderedImage tiles in a > generic way, without making special case for BufferedImage (we have encounter > this issue today in Apache Spatial Information System). > > Regards, > > Martin > > [1] https://bugs.openjdk.java.net/browse/JDK-8166038 > <https://bugs.openjdk.java.net/browse/JDK-8166038> > > <SubimageDrawingTest.java>
