[FFmpeg-devel] [PATCH] Modular test sytem

Stefano Sabatini stefano.sabatini-lala
Sat Dec 26 17:21:57 CET 2009


On date Saturday 2009-12-26 13:17:16 +0100, Stefano Sabatini encoded:
> On date Saturday 2009-12-26 12:46:21 +0100, Diego Biurrun encoded:
> > On Sat, Dec 26, 2009 at 02:01:07AM +0100, Stefano Sabatini wrote:
> > > Sample patch attached, which I tested it with lavf (I created a
> > > tests/lavf.regression.refd/ dir, and I put a file for each test with
> > > the corresponding lines taken from lavf.regression.ref).
> > > 
> > > With the patch applied it is possible to edit the regression.list file
> > > (eliminting or commenting in the corresponding line) for disabling
> > > tests.
> > 
> > I'm not sure I understand the point of this.
> 
> The idea is to be able to disable some test without to edit the
> regression.ref + Makefile.
> 
> With this you only need to eliminate / comment the tests listed in the
> regression.list file and the regression test will run only the
> enabled tests.
> 
> > > --- ffmpeg.orig/Makefile	2009-12-26 01:15:08.000000000 +0100
> > > +++ ffmpeg/Makefile	2009-12-26 02:00:43.000000000 +0100
> > > @@ -156,6 +157,13 @@
> > >  
> > > +$(CODEC_VSYNTH1_REFFILE) = tests/codec.regression.list
> > > +	cat $(patsubst regtest-%,$(SRC_PATH)/tests/codec-vsynth1.regression.refd/%, $(LAVF_TESTS)) > $@
> > > +$(CODEC_VSYNTH2_REFFILE) = tests/codec.regression.list
> > > +	cat $(patsubst regtest-%,$(SRC_PATH)/tests/codec-vsynth2.regression.refd/%, $(LAVF_TESTS)) > $@
> 
> ouch of course that was:
> $(CODEC_VSYNTH1_REFFILE) = tests/codec.regression.list
> 	cat $(patsubst regtest-%,$(SRC_PATH)/tests/codec-vsynth1.regression.refd/%, $(CODEC_TESTS)) > $@
> $(CODEC_VSYNTH2_REFFILE) = tests/codec.regression.list
> 	cat $(patsubst regtest-%,$(SRC_PATH)/tests/codec-vsynth2.regression.refd/%, $(CODEC_TESTS)) > $@
> 
> > > +$(LAVF_REFFILE): tests/lavf.regression.list
> > > +	cat $(patsubst regtest-%,$(SRC_PATH)/tests/lavf.regression.refd/%,          $(LAVF_TESTS)) > $@
> > 
> > Is the ".refd" instead of ".ref" intentional?  If yes, I don't see where
> > that directory is created and am confused.
> 
> Since this patch involves a lot of tedious editing work I didn't
> posted a complete patch, before to do that I want to be sure that the
> idea is going to be accepted.
> 
> The lavf.regression.refd directory needs to contain one file with the
> corresponding test name containing the regression test reference
> output.
> 
> I named it lavf.regression.refd because I don't know how SVN deals
> with this:
> svn remove lavf.regression.ref
> svn mkdir  lavf.regression.ref
> svn add    lavf.regression.ref
> #create tests file ...
> svn add    lavf.regression.ref/{TESTS}
> svn commit lavf.regression.ref
> 
> This has some warts, re-creating the ref files will lose the history
> of the lavf.regression.ref file, on the other hand it will make lot
> easier in the future to check the history of each ref file simply
> doing svn log TEST_REF on the specific file.
> 
> I'll rename lavf.regression.refd -> lavf.regression.ref in a
> successive step.

Anyway complete patch for lavf test attached.

Regards.
-- 
FFmpeg = Fundamental and Frightening Mournful Pacific Enhanced Geisha
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lavf-modular-test-system.patch
Type: text/x-diff
Size: 24464 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20091226/4b4c5ab2/attachment.patch>



More information about the ffmpeg-devel mailing list