Bug 493455 - [win32] fix: source position ignored when drawing transparent image

[jonathan-meier]

Fixes a regression introduced with commit 22797d3 that was reported in Bug 493455 (comment 51).

[iloveeclipse]

Could you add a manual regression test / example?

[jonathan-meier]

I added a JUnit test even, I hope that's fine too. I could easily convert it to a manual test if that's preferred.

[iloveeclipse]

The new test fails... The another one from browser can be ignored.

java.lang.AssertionError: expected:<RGB {0, 255, 0}> but was:<RGB {111, 143, 0}>
	at org.junit.Assert.fail(Assert.java:89)
	at org.junit.Assert.failNotEquals(Assert.java:835)
	at org.junit.Assert.assertEquals(Assert.java:120)
	at org.junit.Assert.assertEquals(Assert.java:146)
	at org.eclipse.swt.tests.junit.Test_org_eclipse_swt_graphics_GC.test_bug493455_drawImageAlpha_srcPos(Test_org_eclipse_swt_graphics_GC.java:681)

[jonathan-meier]

Hmm, I see. I can't reproduce the assertion, neither on my Windows nor on my Linux system, the test passes on both for me. Any hints as to what could be different on your Linux test agent? Otherwise I will just provide a manual test as you suggested.

[SyntevoAlex]

Test fails on Ubuntu 21.04 for me.

I used SwtTestUtil.debugDisplayImage() to see the images and here are srcImage and image:

Note how image has slightly different colors on left and top edges.

I noticed that you copy 100x100 into 200x200, that involves interpolation.
I'm not sure if this is a problem or "working as intended", but anyways, interpolation results in wrong color.

If I alter the test to copy 100x100 to 100x100, test passes.

[jonathan-meier]

Ah yes of course, when stretching the green part, the red part may bleed into the green part at the border due to interpolation. Stretching is actually not necessary so I adjusted the test accordingly and now it passes. Thanks a lot for looking into this @SyntevoAlex!

[SyntevoAlex]

Thanks for fixing it quickly! It's always a pleasure when my effort is not in vain :)

[SyntevoAlex]

The patch makes sense to me and the test now passes.
Review+2 from me.

[iloveeclipse]

@SyntevoAlex : please don't hesitate to merge if you reviewed & approved patches.

[SyntevoAlex]

OK. I wasn't sure if that's an appropriate thing to do now that you already participated.

[iloveeclipse]

test_bug493455_drawImageAlpha_srcPos fails on both MAC platforms

See

• https://download.eclipse.org/eclipse/downloads/drops4/I20220406-1800/testresults/html/org.eclipse.swt.tests_ep424I-unit-mac64-java17_macosx.cocoa.x86_64_17.html
• https://download.eclipse.org/eclipse/downloads/drops4/I20220406-1800/testresults/html/org.eclipse.swt.tests_ep424I-unit-macM1-java17_macosx.cocoa.aarch64_17.html

expected:<RGB {0, 255, 0}> but was:<RGB {255, 255, 255}>

java.lang.AssertionError: expected:<RGB {0, 255, 0}> but was:<RGB {255, 255, 255}>
at org.junit.Assert.fail(Assert.java:89)
at org.junit.Assert.failNotEquals(Assert.java:835)
at org.junit.Assert.assertEquals(Assert.java:120)
at org.junit.Assert.assertEquals(Assert.java:146)
at org.eclipse.swt.tests.junit.Test_org_eclipse_swt_graphics_GC.test_bug493455_drawImageAlpha_srcPos(Test_org_eclipse_swt_graphics_GC.java:681)