Open bastien-roucaries opened 2 months ago
Hi @bastien-roucaries , if you look at the 'golden' images - i.e. the expected output, you can see that it is 17 years old: https://github.com/techblue/jmagick/tree/master/test/correct_output
Imagemagick has probably developed a lot since then and therefore the 'correct_output' has probably changed.
For the failing Images had a difference of (something) which is bigger than max allowed of 20.0:
.... tests, and also the testQuantizeImage_failMightBeOk
, please copy 'actual_output' to 'correct_output' and make a PR and we can look thru the changes and visually check that the acutal output is fine.
The error testDiverseOperations(magicktest.TestJMagick)junit.framework.AssertionFailedError: Scaled to expected:<60> but was:<50>
in https://github.com/techblue/jmagick/blob/master/test/magicktest/TestJMagick.java#L182 sounds strange, like width and height has been interchanged.
You can overwrite the 'golden' testset by setting generate_correct_output = true
in https://github.com/techblue/jmagick/blob/master/test/magicktest/MagickTesttools.java#L35
If you look at https://github.com/techblue/jmagick/blob/master/test/magicktest/MagickTesttools.java#L93 you can see that the comparison is based on mean squared error pixel by pixel (done by invoking compare -metric MSE
), which means that if something has moved slightly in the image, the test will fail. But that might not be an error, but just a change in the underlying ImageMagick implementation, for example due to an improvement of ImageMagick.
Hi
packaging a new version for transitioning to IM7 I get the following test failure
Can you investigate
pipeline here https://salsa.debian.org/java-team/jmagick/-/pipelines/724798