[Pixman] Bill Spitzak patches

Søren Sandmann Pedersen soren.sandmann at gmail.com
Tue Apr 12 02:36:39 UTC 2016


The following patch series contains those of Bill's patches that I
think are ready to be pushed to master, plus some other related
changes that I also think are ready.

01-03: These are patches to do more BILINEAR->NEAREST filter
       reductions. They were inspired by a similar patch in Bill's
       series, but these patches do the reduction in more cases and
       include tests.

04:    Compute the filter size from a transformed ellipse.

05-06: UI fixes to demos/scale.

07:    gnuplot output

       This is based on Bill's gnuplot patch, but I rewrote the logic
       in pixman-filter.c to generate correct coordinates, and added a
       big comment explaining how the phase interleaving works.

08:    Reduce malloc()/free()/memcpy()

09:    Correct Simpson's integration

10:    Integral splitting is only necessary for the LINEAR filter

       I rebased this so that it doesn't depend on the changes to the
       integral() from Bill's series, and made the comment in the code
       match the new code.

11:    Speed up BOX/BOX

       I rebased this and removed the normalization

12:    Fix several issues related to normalization

       This patch fixes several normalization issues including the one
       fixed in Bill's series.

13:    Nested polynomial for cubic

14:    Made Gaussian a bit wider

In the patches where I made changes, I have generally retained Bill as
the author if the patch still contained a substantial block of code
that was written by Bill. Those that I rewrote completely (the
BILINEAR=>NEAREST and the normalization ones), I have put myself as

However, in all cases I'm happy enough to put either me or Bill as the
author. If anyone has strong opinions about this, let me know.

In all the patches I have also reformatted the commit log so that it
fits within 80 characters.

With the exception of the scale->rscale one, I think the remaining
patches in Bill's series should not be accepted, though it is possible
that a new series rebased on top of this will reveal that I
missed something.


More information about the Pixman mailing list