On Wed, 27 Apr 2022 07:11:28 GMT, Sergey Bylokhov <s...@openjdk.org> wrote:
> > The bug is made public. The images looks alike visually. I could not find > > anything wrong with rendering or robot which otherwise could have caused > > issue in other systems and platforms but it's only failing in this CI > > system (but color profile setting is ok). As per > > [JDK-8266247](https://bugs.openjdk.java.net/browse/JDK-8266247) which also > > happens in same mc, some faulty video memory is doubted. Also, it's not > > happening everytime but intermittently so seeing all this, I have just > > relaxed color check a bit > > I am not sure that the color profile or something like that can be the root > cause, because the test does not check an exact RGB of the color, instead, it > renders the image twice and then compares, so it is strange that the results > are different. Not sure I understand. It does not render same image twice but it renders an `ImageIcon `and an `Icon `and compare the bufferedimage of those 2 so I guess it alright to not have exact color check ` if (red1 != red2 || green1 != green2 || blue1 != blue2) {` and rely on color tolerance which we have in many tests involving bufferedimages.. ------------- PR: https://git.openjdk.java.net/jdk/pull/8380
