On Thu, 9 Jun 2022 02:16:45 GMT, Jeremy <d...@openjdk.java.net> wrote:
>> Previously when you drew a line that moved exactly (+0, +0) with a miter >> stroke the Stroker class acted like a horizontal miter was appropriate. Now >> we abort instead, so no miter is applied. >> >> The new unit test tests the original steps plus several variations. >> >> I'm a little nervous about this change because it looks like someone >> previously considered this exact condition and implemented a different >> behavior. I don't think I understand the original rationale. > > Jeremy has updated the pull request incrementally with one additional commit > since the last revision: > > JDK-8264999: GeneralPath.lineTo() to itself produces jagged lines > > Adding "&& prev == DRAWING_OP_TO" per Laurent's code review. > > I also redid the unit tests. Before this change, with the new unit tests: > one scenario failed. With this change: four scenarios failed. > > For now I've commented out the failing tests. The failing tests are not the > same as the original bug; they are related considerations we should also test > against, but IMO they are not essential for the resolution of jdk-8264999. Jeremy, I fixed the bug in the marlin renderer repository, see the Stroker fix: https://github.com/bourgesl/marlin-renderer/commit/b7adf5e606ca0b809118dfe5ab85d4fcf71bd12b#diff-17283d07a3ba6855b0e01eb58cc579c0543b6fae142eb7192dbaa9c9a55e8ef0 Moreover I confirm your 4 disabled tests fail for a good reason: the expected path ignores [...] but the tested path not. In these cases, the expected path (move only) draws nothing, but the tested path (m + ...) will draw something ! Maybe @prrace could confirm ? Here is the passing test on my side: https://github.com/bourgesl/marlin-renderer/commit/b7adf5e606ca0b809118dfe5ab85d4fcf71bd12b#diff-38152f6e0b7e1a267ef1391ec7fea12086ce9b573a55d0bbb7f84bb7eb15170e Please fix the Stroker as proposed, the fix looks good to me (not a reviewer) ------------- PR: https://git.openjdk.org/jdk/pull/8943
