On Tue, 4 Jul 2023 12:53:38 GMT, Alexey Ivanov <aiva...@openjdk.org> wrote:
>> Still I feel that would be blind comparison, since we won't be able to >> exactly test the fix. > > I strongly believe it should be an automated test. It may even be headless, > if possible. There are a couple of examples which compare border rendering. > > You don't need to store images, just count the number of pixels of a > particular colour. If you detect an inconsistency, fail the test. You may > need to tweak the width or height of columns and rows so that old method > results in visible inconsistencies in rendering. How will counting the number of pixels of a particular color help in validating the fix? I was planning to take a portion of image which includes header and 1st row and then check the column of vertical line from header to data, if all are same color then its a straight line, else fail the test... But I was not able to compare the data, some inconsistencies were there with rgb values.... Will try to work on that again...... ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/14766#discussion_r1252525241
