* Update RTD build OS
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Attempt to fix dependencies-latest fail
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Attempt to fix dependencies-latest fail - take 2
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Allow Linux 2023 to use the latest container
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Reorder link libraries
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Fetch matching prebuilt OpenColorIO for OIIO app builds in CI
The aswf/ci-ocio container images intentionally omit OpenColorIO since
the whole point of the image is to build it from source. But the
container's prebuilt OpenImageIO is dynamically linked against a
specific OpenColorIO release that isn't present, causing link
failures for ociolutimage/ocioconvert/ociodisplay whenever
OCIO_USE_OIIO_FOR_APPS=ON. Fetch that matching release from the ASWF
Conan remote and stage it on the linker path before configuring, so
apps link against both the from-source build (used directly) and the
container's expected release (used transitively via OpenImageIO).
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Revert "Reorder link libraries"
This reverts commit fa79c589e0fbb5bb380a5f5406612e48849d9f6a.
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Fail configure when OCIO_USE_OIIO_FOR_APPS is ON but OIIO is missing
Previously, requesting OCIO_USE_OIIO_FOR_APPS=ON while OpenImageIO
could not be found silently fell back to OpenEXR (or skipped building
ociolutimage/ocioconvert/ociodisplay entirely), with only a WARNING.
This let CI silently stop exercising the OIIO code path in those apps
without failing, masking the fact that OIIO wasn't actually being
tested. Turn this into a hard configure-time error instead, so a
missing OpenImageIO is caught immediately when it was explicitly
requested.
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Fix OIIO build on macOS
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Try to identify source of crash in OpOptimizers test
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Print SKIPPED rather than PASSED for new CPU test
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Ensure the new test runs even if the existing test crashes
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
* Turn off AVX512 support for EPYC 9V45
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
---------
Signed-off-by: Doug Walker <doug.walker@autodesk.com>
Doug Walker
c2bd98f7d3