On Fri, 26 Mar 2021 19:53:44 GMT, Phil Race <p...@openjdk.org> wrote:
> https://bugs.openjdk.java.net/browse/JDK-8255800 could have been a one line > spec clean up but > it didn't take a lot of looking to realize there were many more > inconsistencies between spec and implementation. > I've spent a lot of time on what is just small number of factory methods in > Raster because there are so > many possible exceptions and in some cases they rely on other API they call > to generate exceptions and > these may have not been documented or documented acc. to some long lost > behavior. > I've mostly tried to ONLY change spec. But I couldn't help myself when some > checks were missed that > ended up with bizarre and dubious behavior - throwing > NegativeArrayIndexException which just about > always has to be an internal bug ! > > The supplied test passes on JDK 16 as well as this code, because the > (relatively) small number of > cases where JDK 16 threw NegativeArrayIndexException are caught and allowed > only for releases < 17 > So where you see those in the test it corresponds to the behavioral changes > from NegativeArrayIndexException > to IllegalArgumentException. > JCK conformance tests still pass so they must not test those conditions. src/java.desktop/share/classes/java/awt/image/Raster.java line 263: > 261: * both > 0 > 262: * @throws IllegalArgumentException if the product of {@code w} > 263: * and {@code h} is greater than {@code Integer.MAX_VALUE} Do we really want to specify this? I hit this bug during HiDPI work and it seems this is an implementation bug that we do not support big size(w*h). src/java.desktop/share/classes/java/awt/image/Raster.java line 337: > 335: * {@code DataBuffer.TYPE_BYTE}, > 336: * {@code DataBuffer.TYPE_USHORT} > 337: * or {@code DataBuffer.TYPE_INT} What about other types like TYPE_SHORT or we assume it is unsupported? ------------- PR: https://git.openjdk.java.net/jdk/pull/3223