Thanks for the test - it is a good additional data point. If you are able to easily test on Windows, and it fails, that would be very good information to have. It might mean I need to open an issue with OpenJDK.
All of our customers run Windows OS, so that is the only environment we can test with equipment available to us. I have tested with Windows, Java 11, 21, 23 and with 32 and 64 bit variants, and we see the same behavior in all. I've been testing on a Canon LBP6000 (consumer grade laser printer). I don't know the MFP printer model numbers that our customers who have reported the problem have. I am very hesitant to point fingers, but I am beginning to think that the issue is somewhere deep in the Java-Windows Printing native interface. It could even depend on the type of native driver (PCL vs PS vs ???). Side discussion on rasterizing performance: The reason that rasterizing improves performance in our tests is that the print job gets run multiple times with different clipping regions (horizontal stripes at different vertical offsets). Without the rasterization, the print job renders around 20 times. With rasterizing, it renders only twice. This could very well be behavior of Windows printing (maybe to reduce RAM requirements of the print job so they don't have to cache the full page contents before sending to the printer? Who knows). I'm in agreement with Tilman on this - ultimately, printing is such a slow operation that optimizing the Java code is probably not worth the effort or risk of introducing subtle problems. Thanks, K On Sat, Nov 30, 2024, 11:48 AM Tres Finocchiaro <tres.finocchi...@gmail.com> wrote: > > > > So the only way I can demonstrate the failure is to actually print to > > physical paper (you can imagine that this is driving us crazy). I am > > including code to actually run the print jobs at the bottom of this > email. > > > Hi, I'm just a community member but wanted to offer some testing... > > - MFP (consumer grade Epson WorkForce Pro) > - Java 11 > - MacOS arm64 > - All 5 pages of the problematic doc printed fine (tested twice) > > I'd be happy to test on Windows if you believe the bug may be related to > the OS, but I also wanted to see if you've tried to print on other > operating systems, as this may help narrow down the cause. > > My debugging also shows that rasterizing improves printing performance by a > > factor of 10. > > > Just a comment, but this is the first I've heard of rasterizing improving > performance so much. >
