Thanks that looks great, glad to see this stuff is finally in there. I'm disappointed you did not accept the changes to the normalization. I believe I can prove that the residual error will always be zero, unless the filter is so large that it would not work anyway (due to the entries being smaller than the minimum fixed value). You can try the same test I did which led me to this conclusion, try putting a print statement in when the residual error is not zero and see if you can trigger it.
Would like to know if you think there is any chances of getting more changes accepted. I don't want to waste time if that is not going to happen. I would like to move the subsample locations so they are at integers and 1/2, and some other errors, plus push the rejected patches, maybe with better attempts at explaining them. In a longer term I would like to do something to support varying-size filters, and to implement a true 2-pass transform. On Sat, Sep 3, 2016 at 12:10 PM, Søren Sandmann <[email protected]> wrote: > Hi, > > I have pushed these patches to master, except (a) I didn't like the changes > to the normalization patch, so I went with the one I posted back in April > (https://lists.freedesktop.org/archives/pixman/2016-April/004553.html), and > (b) I didn't push the patch adding the filtering dropdown to the demo/scale > program. [1] > > > Søren > > > [1] I accidentally did push it, but reverted it afterwards. > > On Wed, Aug 31, 2016 at 1:03 AM, <[email protected]> wrote: >> >> If anybody is maintaining pixman now, these patches have been reviewed >> several times and should be pushed. They are primarily written by >> Søren based on versions I wrote. They fix several defects in the >> seperable filter generation. >> >> _______________________________________________ >> Pixman mailing list >> [email protected] >> https://lists.freedesktop.org/mailman/listinfo/pixman > > _______________________________________________ Pixman mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/pixman
