Hi, On Mon, Jan 30, 2012 at 2:06 AM, Diego Biurrun <di...@biurrun.de> wrote: > On Sun, Jan 29, 2012 at 01:40:56PM -0800, Ronald S. Bultje wrote: >> On Thu, Jan 26, 2012 at 9:45 AM, Diego Biurrun <di...@biurrun.de> wrote: >> >> > libswscale: >> >> > >> >> > - colorspace-test >> >> > Prints a few lines, then reports success. >> >> >> >> Does it test anything that existing fate tests don't already cover? >> >> >> >> > - swscale-test >> >> > Prints 225k lines of output. >> >> >> >> IIRC the output of swscale-test also depends on the system. >> > >> > Ronald? >> >> swscale-test is system-dependent but it's somewhat useful, it tests >> all scaling/pixfmt combinations to ensure that it works. >> fate-lavfi-pix* only tests with default scaling parameters, so it does >> not test spline/lanczos/bilinear/neighbour/fast-bilinear, only >> bicubic. > > So what about colorspace-test? And how do we make swscale-test useful?
You don't. You instead add proper fate-lavfi-pixfmt_scale_flag variants that test the currently untested variants of scaling. Note, other scaling variants aren't necessarily just different scaling coefficients, some use different implementations of scaling functions. swscale-test is a quick way to test them all, but isn't a proper test until it confirms somehow that the output is meaningful. Ronald _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
