Hi all - I recently did a build of the 10.2 release of oiio, and it seemed
to go smoothly, but there were a lot of test failures.

On comparing the results of testsuite/${testName}/out.txt  and
testsuite/${testName}/ref/out.txt, I found that there were a lot of things
that looked like output format changes that the tests weren't updated for -
ie, in openexr-suite, i get:

oiio:ColorSpace: "Linear"

instead of

Color space: unknown

...all over the place.  However, there are also differences that seem more
ominous, ie:

Comparing "../../../openexr-images-1.5.0/ScanLines/Cannon.exr" and
"Cannon.exr"
Subimage 0: 780 x 566, 3 channel
  Mean error = 0.00227241
  RMS error = 0.00321804
  Peak SNR = 58.308
  Max error  = 0.0874023 @ (487, 6, B)
  441443 pixels (100%) over 1e-06
  441443 pixels (100%) over 1e-06
FAILURE

instead of:

Comparing "../../../openexr-images-1.5.0/ScanLines/Cannon.exr" and
"Cannon.exr"
Subimage 0: 780 x 566, 3 channel
  Mean error = 0
  RMS error = 0
  Peak SNR = inf
  Max error  = 0
  0 pixels (0%) over 1e-06
  0 pixels (0%) over 1e-06
PASS

However, given the fact that it seems the tests are a little out of date,
I'm not sure whether this is something truly meaningful or not.

Has anyone gotten a completely clean test run on a build >= 10.2? Or if not,
know which errors are 'meaningful'?

- Paul
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to