Details
P1: Critical Description
qt3d/tests/auto/threed/qglabstractsurface/qglabstractsurface.pro is marked as an insignificant_test.
Examining recent CI output (last 10 builds) shows that the test passes on Windows (and so should not be insignificant there). On other platforms the test appears to fail in a stable fashion:
linux failure
Testing tst_QGLAbstractSurface FAIL! : tst_QGLAbstractSurface::glWindowSurface() Compared values are not the same Actual (surface1.viewportGL()): QRect(50,50 100x100) (bottomright 149,149) Expected (QRect(0, 0, glw.width(), glw.height())): QRect(0,0 100x100) (bottomright 99,99) Loc: [tst_qglabstractsurface.cpp(100)] FAIL! : tst_QGLAbstractSurface::fboSurface() Compared values are not the same Actual (surface1.viewportGL()): QRect(50,50 100x100) (bottomright 149,149) Expected (QRect(0, 0, glw.width(), glw.height())): QRect(0,0 100x100) (bottomright 99,99) Loc: [tst_qglabstractsurface.cpp(147)] FAIL! : tst_QGLAbstractSurface::subSurface() Compared values are not the same Actual (surface2.viewportGL()): QRect(50,134 32x16) (bottomright 81,149) Expected (QRect(0, glw.height() - 16, 32, 16)): QRect(0,84 32x16) (bottomright 31,99) Loc: [tst_qglabstractsurface.cpp(250)] Totals: 2 passed, 3 failed, 0 skipped
The insignificant_test marker was introduced to mark tests that are unstable, while this test appears to be entirely stable, just failing. This should be corrected by either fixing the failures or marking them as known failures with the QEXPECT_FAIL macro, which would allow the remainder of the test to potentially block future regressions.
Attachments
Issue Links
- depends on
-
QTBUG-24637 tst_QGLAbstractSurface::glWindowSurface failing
-
- Closed
-