pixman-0.43.4-160000.2.2

- Increase test timeout on s390x. Several tests can be slow and
  sometimes times out in our builds.
- Use autosetup macro.

- Update to version 0.43.4
  + Fix incorrect compositing on big-endian architectures.
  + Allow building on clang/arm32.

- a64-neon is an aarch64 feature, not armv6/7

- Rework previous change
  + Disable the demos: pass -Ddemos=disabled to meson. This allows
    us not to BuildRequire gtk3. Having it would result in a build
    cycle.
  + Pass -Dlibpng=disabled to meson: only needed for the tests.
- Drop pkgconfig(gtk+-3.0) BuildRequires: with the demos disabled,
  we don't need this anymore.
- Drop cmake BuildRequires: the build system uses meson. cmake is
  used as a fallback to find libraries when they can't be found
  with pkgconfig.
- Enable test suite (meson_test), without the png tests (for
  minimal build deps).
- Switch post/post to ldconfig_scriptlets: there is no other code
  in the scripts.

- Update to version 0.43.2
  + Highlights include:
  * mmx: use xmmintrin.h if building with SSE2
  * implement r8g8b8 sRGB (without alpha)
  * Build system swapped to Meson
  + Various other fixes
- Add meson BuildRequires and replace configure/make/make_install
  macros with respective meson/meson_build/meson_install, following
  upstreams switch to meson.
- Add cmake and pkgconfig(gtk+-3.0) BuildRequires: new
  depdendencies.

- Update to version 0.42.2 (boo#1205033 CVE-2022-44638):
  + This version contains a fix for a heap overflow.
  + Drop upstream merged pixman-CVE-2022-44638.patch
- Update URL, and tweak source URI.

- Update to version 0.40.0:
  + Meson build system improvements
  + Supporting optimizations on Hygon Dhyana processors
    to match what is enabled for Intel/AMD CPUs.
  + A few Loongson fixes/improvements.
  + Lots of other fixes

- Update to version 0.36.0:
  + Add tests for (a)rgb floating point formats
  + Add support for argb/xrgb float formats, v5
  + Fix stride calculation in stress-test
  + Adjust for clang's removal of __builtin_shuffle
  + Fix vector loads on ppc64le
  + Promote unsigned short to unsigned int explicitly
  + pixman-filter: Made Gaussian a bit wider
  + pixman-filter: Nested polynomial for cubic
  + pixman-filter: Fix several issues related to normalization
  + pixman-filter: Speed up BOX/BOX filter
  + pixman-filter: integral splitting is only needed for triangle
    filterdd
  + pixman-filter: Correct Simpsons integration
  + pixman-filter: reduce amount of malloc/free/memcpy to generate
    filter
  + pixman-image: Added enable-gnuplot config to view filters in
    gnuplot
  + pixman-fast-path.c: Pick NEAREST affine fast paths before
    BILINEAR ones
  + pixman-private: include <float.h> only in C code
- Remove pixman-private-correct-include.patch: upstreamed

- Fix SRPM group. Update summaries.

- Update to GNOME 3.20  Fate#318572

- Add pixman-private-correct-include.patch: pixman-private: include
  <float.h> only in C code, patch from upstream git.

- Update to version 0.34.0:
  + No changes from last version, stable version release only.

- Update to version 0.33.6:
  + configura.ac: fix test for SSE2 & SSSE3 assembler support.
  + mmx: Improve detection of support for "K" constraint.
  + Revert "mmx: Use MMX2 intrinsics from xmmintrin.h directly".

- Update to version 0.33.4:
  + Important security fix to prevent stack buffer overflow.
  + POWER: New fast-paths and optimizations to existing fast-paths.
  + ARMv6: New fast-path.
  + Tests: New cover and fence-image-self tests.
  + Remove redundant corrections in various calculations for
    COVER-CLIP and COVER analysis.

- Update to version 0.33.2:
  + ARMv6 - Many fast paths implementations were added.
  + PPC64/PPC64LE - Fix all outstanding bugs and add many fast
    paths implementations using vmx.
  + Major enhancements to lowlevel-blt-bench utility.
  + A couple of fixes and enhancements to mmx code.

- Update to version 0.32.6:
  + MIPS: Fix exported symbols in public API.
  + build: Check if the compiler supports GCC vector extensions.
  + Bugs fixed: rh#972647, fdo#69014.

- RENDER extension hotfix: Add --disable-vmx for PPC64LE.
  This makes the RENDER extension work for the time being.
  Proper fix for vmx will come later (bnc#876757) (from Egbert
  Eich).

- Update to version 0.32.4:
  + Fix the SSSE3 CPUID detection.
- Changes from version 0.32.2:
  + Build fixes.
- Changes from version 0.32.0:
  + This release contains performance improvements to bilinear
    scaling, and a large number of bug fixes.

- Disable neon on all %arm platforms.

- Update to version 0.30.2:
  + Require GTK+ version >= 2.16.
  + Use AC_LINK_IFELSE to check if the Loongson MMI code can link.
  + configure.ac: Don't use '+=' since it's not POSIX.
  + test:
  - Fix build on MSVC.
  - fix matrix-test on big endian systems.
  + Fix broken build when HAVE_CONFIG_H is undefined, e.g. on
    Win32.

- Update to version 0.30.0:
  + Support for high-quality image downscaling.
  + Much improved ARMv6 assembly.
  + Improved rendering quality for gradients.
  + Improvements to the SSE2 backend.
  + Improvements to MIPS DSPr2 backend.
  + Improvements to test suite.

- Update to version 0.28.2:
  + Fixes for 64 bit Windows, clang, and PowerPC on MacOS and
    OpenBSD.

- Update to version 0.28.0:
  + Support for sRGB coded images.
  + New API for fast glyph rendering.
  + Faster bilinear scaling on iwMMX, Loongson and MMX.
  + More fast paths in the MIPS DSPr2 backend.
  + Faster scaling in general and on SSE2 in particular.

- Update to version 0.26.2:
  + Some bug fixes.
  + Custom build rules for ARM/iwMMXt.
  + An important bug fix for MMX/x86.

- Update to version 0.26.0:
  + New support for MIPS DSPr2 ASE.
  + Improved performance with the iwMMXt instruction set.
  + New support for the  Loongson SIMD instruction set.
  + Several bugs fixed.
  + Cleanups.
  + Performance improvements.

- Fix ARM build:
  + Disable iwmmxt since our ARM SoC does not support it.
  + Add new armv7hl to armv7*

- Parallel build with %_smp_mflags; remove redundant sections

- Update to version 0.24.4:
  + Trapezoids that have gaps between the edges and the top/bottom
    lines are no longer considered invalid.
  + Basic support for using the NEON acceleration on iOS.

- Update to version 0.24.2:
  + Runtime detection of ARM NEON on Android
  + MMX build fix for Solaris
  + Fix for broken gradients with repeat mode NONE
  + Misc. other bug fixes.

- Add pixman-CVE-2022-44638.patch: avoid an integer overflow
  (boo#1205033 CVE-2022-44638).