+++ tests/data/fate/sws-pixdesc-query 2017-10-24 11:25:44.511787161 +0200 @@ -407,8 +407,8 @@ rgb8 rgb4 rgb4_byte - abgr - bgra + argb + rgba rgb48be rgb48le rgb565be @@ -427,8 +427,8 @@ bgr8 bgr4 bgr4_byte - argb - rgba + abgr + bgra bgr565be bgr565le bgr555be Test sws-pixdesc-query failed. Look at tests/data/fate/sws-pixdesc-query.err for details. make: *** [/var/tmp/portage/media-video/ffmpeg-3.3.4/work/ffmpeg-3.3.4/tests/Makefile:220: fate-sws-pixdesc-query] sws-pixdesc-query.err as suggested above shows no further details, it's a 0 Byte file. sws-pixdesc-query.diff shows the same message as above. So I am attatching the build.log only.
Created attachment 499958 [details] emerge --info
Created attachment 499960 [details] build.log
The bug is that this new test makes different assumptions about the code than libswscale, it cannot be reproduced with any command line. Feel free to fix it with this commit: http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=4158fba3cdb7c90f42071323f37f617e4f278414
Same result for media-video/ffmpeg-3.3.6: --- /var/tmp/portage/media-video/ffmpeg-3.3.6/work/ffmpeg-3.3.6/tests/ref/fate/sws-pixdesc-query 2017-12-31 23:35:50.000000000 +0100 +++ tests/data/fate/sws-pixdesc-query 2018-01-20 14:28:49.055068536 +0100 @@ -407,8 +407,8 @@ rgb8 rgb4 rgb4_byte - abgr - bgra + argb + rgba rgb48be rgb48le rgb565be @@ -427,8 +427,8 @@ bgr8 bgr4 bgr4_byte - argb - rgba + abgr + bgra bgr565be bgr565le bgr555be Test sws-pixdesc-query failed. Look at tests/data/fate/sws-pixdesc-query.err for details. make: *** [/var/tmp/portage/media-video/ffmpeg-3.3.6/work/ffmpeg-3.3.6/tests/Makefile:220: fate-sws-pixdesc-query] Error 1
As said before, the bug is in the test itself, there is no (related) bug in the tested code.
Ok for me. Also I am not demanding anything. But a buggy test-procedure could be seen as a bug too, even if the tested code works. ;-) Maybe I should take this upstream if media-video/ffmpeg-3.4.x is affected in the same way.
(In reply to ernsteiswuerfel from comment #6) > Ok for me. Also I am not demanding anything. But a buggy test-procedure > could be seen as a bug too, even if the tested code works. ;-) I don't disagree, just suggesting that you remove the test as it is buggy. > Maybe I should take this upstream if media-video/ffmpeg-3.4.x is affected in > the same way. https://trac.ffmpeg.org/ticket/6554 (the hurd issue is unrelated).
Fixed in media-video/ffmpeg-3.4.2. It passes all tests on ppc now. So the fix Carl was talking about must have trickled down to 3.4.x. See: http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=4158fba3cdb7c90f42071323f37f617e4f278414
*** Bug 650206 has been marked as a duplicate of this bug. ***
