[FFmpeg-devel] Revision 9087 regression test fail

Benoit Fouet benoit.fouet
Tue May 29 12:42:39 CEST 2007


Diego Biurrun wrote:
> On Tue, May 29, 2007 at 11:23:03AM +0200, Benoit Fouet wrote:
>   
>> Diego Biurrun wrote:
>>     
>>> On Wed, May 23, 2007 at 12:59:59PM +0200, Benoit Fouet wrote:
>>>   
>>>       
>>>> Michael Niedermayer wrote:
>>>>
>>>>         
>>>>>>>> Here is a starting part of the regression test report where errors
>>>>>>>> initially pop up (the rest of log is attached):
>>>>>>>>             
>>>>>>>>                 
>>>>>>> the attached file is definitly not the complete output of "make test"
>>>>>>> i think its not even the complete output of make seektest as that
>>>>>>> should output a proper diff but the headers are missing ...
>>>>>>>           
>>>>>>>               
>>>>>> Yes, indeed, it only contained the last part of log with errors (the same
>>>>>> information which was enough for you in the report of Panagiotis Issaris).
>>>>>>
>>>>>> I'm interpreting your reply as a request for complete output of "make test".
>>>>>> Now attached the file with the output generated in 'maketest.log' by invoking
>>>>>> the following command from shell: "make test &> maketest.log"
>>>>>>         
>>>>>>             
>>>>> does it work with --enable-gpl? (it should)
>>>>>       
>>>>>           
>>>> it does (at least on my machine)
>>>>         
>>> But is it the correct fix?  Why does it depend on --enable-gpl?
>>>       
>> i haven't looked further to know what make the dependance necessary.
>> but without, it fails on some machine (e.g. on mine :) )
>>     
>
> I think the problem is a different one.  You need to run codectest
> before seektest.  Does it work if you do that?
>
>   

no, i keep on having the same issue:
ls: cannot access data/b-*: No such file or directory

>>>>> if so the seek test should be disabled without that ...
>>>>>       
>>>>>           
>>>> a patch trial is attached
>>>>
>>>> --- tests/Makefile	(revision 9103)
>>>> +++ tests/Makefile	(working copy)
>>>> @@ -52,8 +52,10 @@
>>>>  	@exit 1
>>>>  endif
>>>>  
>>>> +ifeq ($(CONFIG_GPL),yes)
>>>>  seektest: seek_test$(EXESUF)
>>>>  	@$(SRC_DIR)/seek_test.sh $(SEEK_REFFILE)
>>>> +endif
>>>>  
>>>> @@ -85,8 +87,10 @@
>>>>  
>>>>  #FIXME cleanup shit below
>>>> +ifeq ($(CONFIG_GPL),yes)
>>>>  seek_test$(EXESUF): seek_test.c $(LIBAV)
>>>>  	$(CC) $(LDFLAGS) $(CFLAGS) -DHAVE_AV_CONFIG_H -I.. -I$(SRC_PATH)/libavformat -I$(SRC_PATH)/libavcodec -I$(SRC_PATH)/libavutil -o $@ $< $(SRC_PATH)/libavformat/libavformat.a $(SRC_PATH)/libavcodec/libavcodec.a $(SRC_PATH)/libavutil/libavutil.a $(EXTRALIBS)
>>>> +endif
>>>>         
>>> This is very ugly, there is no need to add two more conditionals.
>>>       
>> i'm not very familiar with makefile stuff, what is the correct way to do
>> it ?
>>     
>
> The second conditional is superfluous and the seektest rule can be moved
> inside the conditional that is already present.
>   

how is the attached patch ?

Ben
-- 
Purple Labs S.A.
www.purplelabs.com
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: seek.diff
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070529/18301ffa/attachment.txt>



More information about the ffmpeg-devel mailing list