On Fri, Mar 13, 2009 at 2:38 PM, Henrik Akesson <h.m.akesson@xxxxxxxxx> wrote:
> I'd like to set up some regression tests before doing any changes to the
> code.
>
> You agree on using the glib test framework?
>
> I'll do it by applying some operations on an input image file and comparing
> the resulting output image file with a reference. However, I suppose that
> this is not platform independent. There would be slight differences in the
> precision of the calculations, resulting in slightly different pixel values
> and files that are not at all comparable.
>
> Any ideas/suggestions?
The gallery in docs/ could be extended to also store reference images.
I've been using this setup to look for rendering and performance
regressions (it used to even be built and have a browsable
multi-architecture continous build and testing at Open Source
Development Labs but that effort was shut down with some of OSDLs
cut-backs in Engineering operations).
There is a set of regression tests for GeglBuffer in tests/buffer
which outputs grayscale images as unicode and uses unified diffs to
detects differences, I think Martin Nordholts has mentioned that he
wants to port these tests to use glib test framework.
/Øyvind K.
--
«The future is already here. It's just not very evenly distributed»
-- William Gibson
http://pippin.gimp.org/ http://ffii.org/
_______________________________________________
Gegl-developer mailing list
Gegl-developer@xxxxxxxxxxxxxxxxxxxxxx
https://lists.XCF.Berkeley.EDU/mailman/listinfo/gegl-developer
