On 04/07/2015 12:19 PM, Matt Turner wrote:
I don't know how Cairo does review, but I think it would be really nice if a Cairo developer reviewed Bill's patches (I think they were adding a new API to pixman?) if not for all the little technical details but that the API makes sense for its uses in Cairo.
My patches are to move the current cairo FILTER_GOOD/BEST behavior into pixman. Pixman can do this better and faster, and it means that the X11 Cairo backend does not have to do an image fallback for transforms.
The main difference between this patch and the Cairo code is that Søren wanted to preserve his current api for generating filters, which requires selection of two filters which are then integrated (one of them scaled by the width). I am not in full agreement with this as all other systems I have ever seen treat the smaller filter as impulse, and except for box+box (which produces a trapezoid and most software produces that directly when "box" is requested), I have to use impulse for one of the filters always.
In any case most of the series is to fix bugs and inefficiencies in the filter generation.
There is then a patch to enable use of the filtering for GOOD/BEST settings, matching Cairo (with a few minor improvements).
There is also patches to fix the demo programs so the GOOD/BEST filter can be tested and to fix the filter size for rotated images.
In addition if this is pushed we need a test in Cairo to detect if pixman is new enough and then remove Cairo's local emulation. There has to be a test for the linked Pixman and perhaps a different one for the X server.
_______________________________________________ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman
