Andrea,I saw a single rendering test failure on Debian unstable with OpenJDK 8 so I increased the pixel difference tolerance in renderer VectorFillTest.testCrossFill. This test was introduced here:
https://github.com/geotools/geotools/commit/42c4cc0c3d850b1dea50c07108f171aace35a798 https://osgeo-org.atlassian.net/browse/GEOT-5381
With so many sharp edges (a fill with hollow crosses) this test is sensitive to slight antialiasing differences. I saw 196 different pixels which exceeds the limit of 100, so I increased the limit to 300 for this test only. This is still pretty strong agreement for a 300x300=90000 pixel image. I have attached the images and diffs. I applied the fix to master, 14.x, and 13.x. Please let me know if you have any objection.
If we get reports of failures in other test methods of VectorFillTest on some platforms, it may be better to increase the default tolerance for the whole fixture.
Nice work by the way. Renderer tests are so pretty. Kind regards, -- Ben Caradoc-Davies <b...@transient.nz> Director Transient Software Limited <http://transient.nz/> New Zealand
------------------------------------------------------------------------------ Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785231&iu=/4140
_______________________________________________ GeoTools-Devel mailing list GeoTools-Devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-devel
