diff options
| author | Roman Scherer <roman@burningswell.com> | 2025-12-14 19:23:13 +0100 |
|---|---|---|
| committer | Andreas Enge <andreas@enge.fr> | 2025-12-16 20:28:09 +0100 |
| commit | 1cc99c8565af9c56bcc93d2635704767465e3cc2 (patch) | |
| tree | 647ecb8bdf8d6dda2f5ce24d954cccb54c2b8421 /gnu/packages/image-processing.scm | |
| parent | c1f2a0e477b31dfac3a207fa6499ac6ea4a2ddd5 (diff) | |
gnu: opencv: Skip failing test on aarch64.
* gnu/packages/image-processing.scm (opencv)[arguments]: Disable
ReferenceAccuracy test due to floating-point precision differences
with ARM NEON.
Signed-off-by: Andreas Enge <andreas@enge.fr>
Diffstat (limited to 'gnu/packages/image-processing.scm')
| -rw-r--r-- | gnu/packages/image-processing.scm | 8 |
1 files changed, 7 insertions, 1 deletions
diff --git a/gnu/packages/image-processing.scm b/gnu/packages/image-processing.scm index 9d5cd62144b..d6b490980cc 100644 --- a/gnu/packages/image-processing.scm +++ b/gnu/packages/image-processing.scm @@ -645,7 +645,13 @@ different data arrays similar to those available in the numdiff software.") (substitute* "modules/photo/test/test_hdr.cpp" (("0\\.131") "0.222")) ;; These tests hang forever on aarch64. - (delete-file-recursively "modules/videoio/test/")) + (delete-file-recursively "modules/videoio/test/") + ;; This test fails on aarch64 due to floating-point precision + ;; differences with ARM NEON. + ;; Expected: RMSE <= 0.34, actual: 0.407627 + (substitute* "../opencv-contrib/modules/optflow/test/test_OF_accuracy.cpp" + (("\\bReferenceAccuracy\\b" all) + (string-append "DISABLED_" all)))) '()) ,@(if (target-riscv64?) |