Re: convwatch.py

Am 03.07.12 14:51, schrieb Thorsten Behrens:
> Peter Jentsch wrote:
> > I've noticed your recent revival of convwatch.py for layout regression
> > tests. We've been developing a tool (written in Java) that might make
> > convwatch.py more portable, by allowing to ignore certain graphical
> > differences (caused by the possibly different implementations of
> > antialiasing on windows and linux for example). We'd like to integrate
> > that with convwatch.py to make it part of the LibO regression tests
> > *but* I'm quite unsure about how you're using convwatch.py currently.
> > Could you help us out with a hint here?
> Hi Peter,
>
> all image comparison scripts we have ultimately call imagemagick's
> composite, followed by pixel counting via identify - so having a
> stand-alone command line version of your smart comparison would be 
> perfect I guess. :)
>
> Cheers,
>
> -- Thorsten

Hi Thorsten,

thanks for your quick reply. We'll just try our tool combined with
convwatch on some of the testdocuments in the LibO source tree and see
what happends.

Cheers,

Peter