AWT integration tests: Native: False positives on systems with older libs (fontconfig ?)

[Karm]

Describe the bug

I came across JDK x Mandrel portable builds (regardless of Mandrel version, both 21.2, 21.3 and 22.0) that pass Quarkus AWT tests on RHEL 9 ✔️ , but fail on RHEL 8 ❌ :

TIFF: Wrong pixel at 275x199. Expected: [255,0,0,255] Actual: [171,37,162,255] ==> expected: <true> but was: <false>
PNG:  Wrong pixel at 275x199. Expected: [255,0,0,255] Actual: [171,37,162,255] ==> expected: <true> but was: <false>
JPG:  Wrong pixel at 275x199. Expected: [254,2,0,0]   Actual: [174,36,157,0]   ==> expected: <true> but was: <false>
BMP:  Wrong pixel at 275x199. Expected: [255,0,0,0]   Actual: [171,37,162,0]   ==> expected: <true> but was: <false>

The test was looking for an edge of a character rendered with a specific font.

This might come down to a different native library version being used (fonts?) and needs attention.

I will either attend to the issue on the Mandrel side or adjust the AWT test suite to be more forgiving than it already is with thresholding expected pixel values.

Expected behavior

No response

Actual behavior

No response

How to Reproduce?

No response

Output of uname -a or ver

No response

Output of java -version

No response

GraalVM version (if different from Java)

No response

Quarkus version or git rev

No response

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[quarkus-bot]

/cc @galderz, @zakkak

[Karm]

The animation shows the difference. The two top left texts are rendered with font embedded in the test suite whereas the rest of the text is rendered using default fonts either brought with the JDK or present in the system. I cannot reproduce it on a stock, clean install of RHEL 8.5. It looks the same as it does on modern Fedora, Centos 8 Stream, Ubuntu (including GitHub Actions instances). It is just a particular RHEL 8 setup that renders a different image. I will dig deeper and modify the TS to run everywhere as this is a false positive.

@zakkak FYI to see the nice picture :-D

[Karm]

The root cause was different system fonts, not libraries. Solved in #22755