[cairo] cairo snapshot 1.11.4 now available

Dimiter 'malkia' Stanev malkia at gmail.com
Mon Mar 12 11:57:10 PDT 2012

Hi Chris,

This release works fine for me on OSX (10.5+), and probably would on 
Linux. It still does not work on Windows (the .svg/.pdf, and probably 
other vector backends work fine, but PNG and Win32 does not).

It actually works, up to the point where cairo_show_text is used.
I've tried building it with and without FreeType - same result.

The last version that worked for me was this:

Author: Adrian Johnson <ajohnson at redneon.com>  2011-09-10 07:09:15
Committer: Adrian Johnson <ajohnson at redneon.com>  2011-09-10 08:15:09
Parent: 2d887d99c74f2097021a1f4f7c1d2306b8f9b429 (ps: use reusable 
stream for mesh pattern data)
Child:  0000000000000000000000000000000000000000 (Local uncommitted 
changes, not checked in to index)
Branches: master, remotes/origin/master
Follows: 1.11.2
Precedes: 1.11.4
     ps: improve formatting of fallback image comment

The next one following it (by what I see it in gitk/gitx, not 
time-wise) that does not work is this:

Author: Chris Wilson <chris at chris-wilson.co.uk>  2011-07-30 09:28:21
Committer: Chris Wilson <chris at chris-wilson.co.uk>  2011-09-12 00:29:48
Parent: 0540bf384aed344899417d3b0313bd6704679c1c (ps: improve formatting 
of fallback image comment)
Child:  65a954d5bab9ab6fed15bd98b7018aca2fc50107 (test-surfaces: 
compilation fixes)
Branches: master, remotes/origin/master
Follows: for-ufo
Precedes: 1.11.4

It's not stopping my work, as cairo is so awesome that the ABI did not 
changed to affect me (which makes me smile, as this is not often the 
case in the industry where I'm in).

Also I see that this is just snapshot, so not full release, but just in 
case if this would be needed, here is the feature list I've compiled it 



On 3/12/2012 3:18 AM, Chris Wilson wrote:
> A new cairo snapshot 1.11.4 is now available from:
> 	http://cairographics.org/snapshots/cairo-1.11.4.tar.gz
>      which can be verified with:
> 	http://cairographics.org/snapshots/cairo-1.11.4.tar.gz.sha1
> 	ac4c37de8827ec1c4092509b9eec72106173d350  cairo-1.11.4.tar.gz
> 	http://cairographics.org/snapshots/cairo-1.11.4.tar.gz.sha1.asc
> 	(signed by Chris Wilson)
>    Additionally, a git clone of the source tree:
> 	git clone git://git.cairographics.org/git/cairo
>      will include a signed 1.11.4 tag which points to a commit named:
> 	a8dbc68056af9f32b6ebc6c1a0191e9ca7ec2e7d
>      which can be verified with:
> 	git verify-tag 1.11.4
>      and can be checked out with a command such as:
> 	git checkout -b build 1.11.4
> Snapshot 1.11.4 (2012-13-12)
> ============================
> The cairo community is pleased to finally announce the long aniticpated
> release candidate for 1.12, 1.11.4, of the cairo graphics library. This
> is the first major update to cairo in over a year and brings a large
> number of new features; undoubtably a few bugs as well.
> While many people have contributed and have helped to test the release,
> providing feedback on 1.10 and suggesting improvements, this release
> is the result of a few persevering souls who deserve recognition for their
> outstanding contributions: Andrea Canciani (all round bug fixing,
> performance tuning and master of the gradients), Adrian Johnson (PDF
> supremo) and Uli Schlachter (who stepped forward as maintainer for the
> XCB backend).
> Major additions since 1.11.2:
>   * cairo_surface_map_to_image API for pixel level access to any surface
>   * New antialias hints to control the trade-off between speed and
>   * quality
>   * A callback pattern, cairo_pattern_create_raster_source, for lazy
>     decoding of image data.
>   * cairo_surface_observer_t, a new type of surface to gather performance
>     statistics
>   * XCB as a supported backend
>   * A rewritten compositor pipeline for performance improvements for, but not
>     limited to, the xlib and image backends.
>     From ION and PineView through to SandyBridge, every machine I have shows
>     across the board performance improvement on the cairo-traces:
>     i5-2520m	gnome-system-monitor:	5.97x speedup
>     pnv		gnome-system-monitor:	4.86x speedup
>     i5-2520m	firefox-asteroids:	4.66x speedup
>     pnv		firefox-asteroids:	4.43x speedup
>     image	firefox-canvas:		3.82x speedup
>     i5-2520m	firefox-canvas-alpha:	3.49x speedup
>     image	firefox-asteroids:	2.87x speedup
>     pnv		firefox-talos-svg:	2.83x speedup
>     ion		grads-heat-map:		2.75x speedup
>     pnv		firefox-canvas-alpha:	2.66x speedup
>     image	gnome-system-monitor:	2.66x speedup
>     image	swfdec-giant-steps:	2.46x speedup
>     image	firefox-canvas-alpha:	2.14x speedup
>     i5-2520m	firefox-talos-svg:	2.03x speedup
>     image	grads-heat-map:		2.02x speedup
>     ion		gnome-system-monitor:	2.00x speedup
>     pnv		firefox-particles:	1.99x speedup
>     i5-2520m	grads-heat-map:		1.96x speedup
>     pnv		firefox-canvas:		1.92x speedup
>     ion		firefox-particles:	1.80x speedup
>     image	poppler-reseau:		1.77x speedup
>     pnv		xfce4-terminal-a1:	1.72x speedup
>     image	firefox-talos-svg:	1.65x speedup
>     pnv		grads-heat-map:		1.63x speedup
>     i5-2520m	firefox-canvas:		1.63x speedup
>     pnv		swfdec-youtube:		1.62x speedup
>     image	ocitysmap:		1.59x speedup
>     i5-2520m	firefox-fishbowl:	1.56x speedup
>     i5-2520m	poppler-reseau:		1.50x speedup
>     i5-2520m	evolution:		1.50x speedup
>     i5-2520m	midori-zoomed:		1.43x speedup
>     pnv		firefox-planet-gnome:	1.42x speedup
>     i5-2520m	firefox-talos-gfx:	1.41x speedup
>     i5-2520m	gvim:			1.41x speedup
>     pnv		ocitysmap:		1.37x speedup
>     image	poppler:		1.31x speedup
>     ion		firefox-canvas-alpha:	1.35x speedup
>     ion          firefox-talos-svg:	1.34x speedup
>     i5-2520m	ocitysmap:		1.32x speedup
>     pnv		poppler-reseau:		1.31x speedup
>     i5-2520m	firefox-planet-gnome:	1.31x speedup
>     pnv		firefox-fishbowl:	1.30x speedup
>     pnv		evolution:		1.28x speedup
>     image	gvim:			1.27x speedup
>     i5-2520m	swfdec-youtube:		1.25x speedup
>     pnv		gnome-terminal-vim:	1.27x speedup
>     pnv		gvim:			1.25x speedup
>     image	firefox-planet-gnome:	1.25x speedup
>     image	swfdec-youtube:		1.25x speedup
>     ...
> And a plethora of minor improvements everywhere!
> -Chris
> What is cairo
> =============
> Cairo is a 2D graphics library with support for multiple output
> devices. Currently supported output targets include the X Window
> System, quartz, win32, and image buffers, as well as PDF, PostScript,
> and SVG file output. Experimental backends include OpenGL, XCB, BeOS,
> OS/2, and DirectFB.
> Cairo is designed to produce consistent output on all output media
> while taking advantage of display hardware acceleration when available
> (for example, through the X Render Extension).
> The cairo API provides operations similar to the drawing operators of
> PostScript and PDF. Operations in cairo include stroking and filling
> cubic Bézier splines, transforming and compositing translucent images,
> and antialiased text rendering. All drawing operations can be
> transformed by any affine transformation (scale, rotation, shear,
> etc.).
> Cairo has been designed to let you draw anything you want in a modern
> 2D graphical user interface.  At the same time, the cairo API has been
> designed to be as fun and easy to learn as possible. If you're not
> having fun while programming with cairo, then we have failed
> somewhere---let us know and we'll try to fix it next time around.
> Cairo is free software and is available to be redistributed and/or
> modified under the terms of either the GNU Lesser General Public
> License (LGPL) version 2.1 or the Mozilla Public License (MPL) version
> 1.1.
> Where to get more information about cairo
> =========================================
> The primary source of information about cairo is:
> 	http://cairographics.org/
> The latest versions of cairo can always be found at:
> 	http://cairographics.org/download
> Documentation on using cairo and frequently-asked questions:
> 	http://cairographics.org/documentation
> 	http://cairographics.org/FAQ
> Mailing lists for contacting cairo users and developers:
> 	http://cairographics.org/lists
> Roadmap and unscheduled things to do, (please feel free to help out):
> 	http://cairographics.org/roadmap
> 	http://cairographics.org/todo
> List of all changes since 1.11.2
> ================================
> Adrian Johnson (90):
>        Type1-subset: Fallback if font contains more than one /Encoding
>        cff: Fix heap corruption
>        PS: Use tight bounding box
>        png: fix type1 encoding bugs
>        cff-subset: Fix /guillemotright SID encoding
>        type1-subset: Remove FT dependency part 1
>        type1-subset: Remove FT dependency part 2
>        type1-subset: Remove FT dependency part 3
>        type1-subset: Remove FT dependency part 4
>        type1-subset: Remove FT dependency part 5
>        type1-subset: Remove FT dependency part 6
>        Add win32 font backend functions for Type 1 subsetting
>        Remove HAS_FT_FONT from cairo-scaled-font-subsets-private.h
>        win32: Fix build breakage
>        ps: Don't return CAIRO_INT_STATUS_NOTHING_TO_DO for empty glyphs
>        analysis: CAIRO_INT_STATUS_NOTHING_TO_DO should not cause fallback
>        Add support for subsetting bare CFF fonts
>        type1-subset: Subset the subroutines
>        Check glyph_index range in _index_to_glyph_name
>        ps: remove unnecessary flush
>        pdf: check if images are grayscale or monochrome and encode as such
>        image: fix bug in analyze_color
>        ps: check if images are grayscale or monochrome and encode as such
>        image: check if A8 image is bilevel when analyzing transparency
>        pdf: check if smask is bilevel and encode as such
>        ps: use show_text_glyphs
>        subsets: use show_text_glyphs supplied utf8 to determine latin character
>        ps: use deflate compression for ps level 3
>        pdf: change end of line in pdf output from \r\n to \n
>        cff-subset: fallback when parsing the charstrings in bare cff fonts fails
>        cff-subset: don't easily give up parsing a charstring if we already have the width
>        cff-subset: fix decoding of real numbers
>        font-subsets: fix wrong string length
>        ps/pdf: use a new clipper object when emitting a recording surface
>        pdf: use a new clipper object in recording_subsurface
>        pdf: use ink extents for smask bbox
>        pdf: use ink extents for pattern bbox
>        pdf: use ink extents in content stream XObjects
>        pdf: each row of 1-bit image data needs to begin on a byte boundary
>        ps: each row of 1-bit image data needs to begin on a byte boundary
>        pdf: use _emit_smask() instead of _emit_imagemask to emit stencil mask
>        ps: support stencil masks
>        ps/pdf: fix the polarity of stencil masks
>        pdf: use interpolate in image smask
>        cff: strip subset tag when reading font name
>        pdf: latin text must use character codes in the toUnicode CMap, not glyph indices
>        ps: use reusable stream for mesh pattern data
>        ps: improve formatting of fallback image comment
>        pdf: add status check
>        ft: fix warning
>        scaled-subsets: ensure different glyphs with same utf8 mapping are not merged together
>        ps: turn off debug
>        Refresh some pdf/ps ref images
>        subsetting: Support unicode fontnames
>        analysis mask: unwrap recording surface snapshot
>        test: add mask-alpha ps ref
>        test: refresh some pdf/ps refs
>        test: refresh some pdf/ps refs
>        pdf: ensure ink extents is not larger than bounded extents
>        test: refresh pdf mask ref images
>        subsets: latin char map needs to belong to the font, not the subset
>        test: Add stroke-pattern
>        test: make text-pattern draw with both opaque and translucent patterns
>        cff: in CID fonts the CID is the glyph index
>        cff: fallback if seac style endchar is found
>        truetype-subset: remove unused variable
>        type1-subset: remove unused variables
>        pdf: don't use patterns with padded images
>        ps: avoid padding images if the padding is not required to fill the extents
>        pdf: allow embedding of cmyk jpeg images
>        ps: allow embedding of cmyk jpeg images
>        pdf: avoid using pdf patterns to paint/fill opaque linear/radial gradients
>        pdf: avoid using pdf patterns to paint/fill translucent linear/radial gradients
>        ps: simplify the EPS save and restore
>        doc: fix typos
>        test: add fallback
>        pdf: support RASTER_SOURCE patterns
>        test: add push/pop group to fallback test
>        ps: handle different x/y fallback resolution in Fallback Image comment
>        api: add cairo_surface_supports_mime_type
>        ps: support RASTER_SOURCE patterns
>        ps: avoid using ps patterns to paint/fill gradients
>        ps: add missing 'Q' to end of page
>        ps: ensure shading domain is [ 0 1 ]
>        ps: fix extend-pad-border test failure
>        type1-subset: some fonts have "noaccess put" instead of NP in the Subrs
>        type1-subset: ensure additional glyphs required by seac operator are parsed
>        win32: fix compilation on cygwin
>        test: add test to check that pdf transparency groups are non-isolated
>        pdf: ensure all transparency groups are isolated
> Alexandros Frantzis (32):
>        gl: Fail if GL implementation doesn't support shaders
>        gl: Remove unnecessary checks for NULL shader implementation
>        gl: Remove fixed-function related code paths
>        gl: Add entry for UniformMatrix4fv in the gl dispatch table
>        gl: Add function to bind a 4x4 float matrix shader uniform
>        gl: Replace ftransform() with manual coordinate calculation in shaders
>        gl: Use a custom shader uniform for the ModelViewProjection matrix
>        gl: Add gl dispatch entries for functions related to vertex attributes
>        gl: Replace built-in vertex attributes with custom attributes
>        egl: Ensure that the dummy pbuffer surface is compatible with the supplied context
>        boilerplate/egl: Ensure that we are using an RGBA32 egl config
>        gl: Correctly extract GL version from OpenGL ES version strings
>        gl: Store the GL flavor (Desktop or ES) in the device struct
>        gl: Add GLES2 support to dispatch table initialization
>        gl: Pass more information to the shader emitter functions
>        gl: Add functions to get the filter and extend of gl operands.
>        gl: Provide a shader implementation of GL_CLAMP_TO_BORDER for GLES2
>        gl: Store gradients in 2D textures of height 1 instead of 1D textures
>        Extract function to query endianness from xlib/xcb and make it common
>        gl: Require the GL_EXT_texture_format_BGRA8888 extension for GLES2
>        gl: Use conventional texture upload for GLES2
>        gl: Add support for creating image surfaces for GLES2
>        gl: Add GLES2 support for drawing image surfaces on gl surfaces
>        gl: Remove unnecessary call to glDisableClientState()
>        gl: Use GL_MAX_TEXTURE_IMAGE_UNITS to query the maximum texture units
>        gl: Require the GL_OES_texture_npot extension for GLES2
>        gl: Add GLESv2 backend to build system
>        gl: Fix build issues for GLES2
>        boilerplate/egl: Add GLES2 support
>        gl: Define the float precision in the fragment shader for GLES2
>        gl: Fall back to dlsym() if *GetProcAddress() fails
>        gl: Add fallback path for GLES2 implementations not supporting GL_OES_mapbuffer
> Andrea Canciani (189):
>        Fix regressions from 1.10 merge
>        xcb: Fix compositing of pixel-aligned rectangles
>        mesh: Rename cairo_pattern_mesh_* functions to cairo_mesh_pattern_*
>        test: Fix clip-device-offset
>        test: Fix surface-pattern-scale-down-extend
>        test: Fix REFERENCE_IMAGES
>        xlib: Fix compilation when gradient functions are not available
>        image: Fix surface acquisition if PIXMAN_HAS_ATOMIC_OPS
>        test: Fix missing REFERENCE_IMAGES
>        test: Silence gcc
>        test: Silence gcc warning
>        test: Silence gcc warnings
>        test: Cleanup macros
>        test: Use ARRAY_LENGTH() macro
>        test: Add show-glyphs-advance
>        quartz: Fix y glyph advance
>        xcb,xlib: Fix 1-stop gradients
>        xcb,xlib: Fallback upon generic radial gradients
>        test: Add stroke-open-box
>        path: Fix _cairo_path_fixed_is_rectangle()
>        script: Fix stroking of 3 sided boxes
>        Make cairo_set_font_size() a wrapper of cairo_set_font_matrix()
>        quartz-font: Remove unused variables
>        Improve cairo_rectangle_list_t error handling
>        trace: Always emit a function name for write-to-png-stream
>        image: Silence missing enum value warning
>        image: Silence gcc warning
>        Do not warn when ignoring the return value of _cairo_rectangle_intersect()
>        Remove unused function declarations
>        Do not use fmax()
>        Revert "Add an implementation of fmax for use with MSVC"
>        Improve headers-standalone target
>        Fix warning by check-doc-syntax
>        Fix visibility issues found by make check
>        Remove if's from Makefile.sources
>        boilerplate: Add generation of constructors on win32
>        test: Update generation of constructors on win32
>        test: Fix compilation on win32
>        test: Use POSIX-compatible unlink
>        path-stroke: Avoid mixing cap and join values
>        toy-font-face: Do not assume hash != 0
>        xlib: Fix build of xlib-xcb
>        build: Don't build cairo-sphinx when the tee surface is off
>        xlib-xcb: Implement cairo_xlib_device_debug_{g,s}et_precision
>        script: Return status when flushing device
>        toy-font: Do not open-code cairo_font_face_reference ()
>        boilerplate: Silence MSVC warnings
>        Avoid deprecated functions on win32
>        win32-font: Do not return font in error status
>        win32-font: Implement destroy function
>        win32-font: Improve static data reset function
>        build: Improve generation of cairo-features.h on win32
>        build: Improve dependencies in win32 Makefiles
>        build: Use common rules for making pdiff
>        build: Clean up environment variables and flags for the win32 build
>        build: Silence autoconf warnings
>        Hide private symbols on MacOSX
>        error: Do not define _cairo_error twice
>        test: Cleanup build on Win32
>        script-interpreter: Build on win32
>        perf: Ensure M_SQRT2 is defined
>        perf: Fix build on win32
>        test: Improve check-ref-missing
>        test: Fix typo
>        clip: Fix boxes extents computation in intersect_with_boxes
>        test: Make parsing of log files more solid
>        test: Ignore \r when parsing test log files
>        test: Add extended-blend-solid tests
>        win32: Fallback upon PDF blend operators
>        test: Merge text-antialias-*
>        test: Fix check-ref-missing
>        Do not open-code freed-pool
>        Remove conditional compilation of freed-pools
>        Clean up context_pool upon static data reset
>        Make error contexts static
>        pattern: Implement _rgb functions as wrappers over _rgba functions
>        color: Remove unused functions
>        gstate: Remove useless code
>        device: Define the CAIRO_DEVICE_TYPE_INVALID device type
>        device: Fix documentation of device types
>        build: Fix automake error
>        build: Fix compilation
>        xcb,xlib: Cleanup GC cache handling
>        Use new pixman formats
>        quartz: Silence const cast warnings
>        Fix surface backend structures
>        win32: Fix compilation
>        win32: Constify
>        surface: Check image parameter in cairo_surface_unmap_image()
>        win32: Fix typos in documentation
>        test: Update image references
>        test: Improve horizontal-clip
>        polygon: Simplify code for edge clipping
>        polygon: Fix generic case of edge clipping
>        script: Initialize recording extents
>        directfb: Unbreak compilation
>        Silence gcc warnings
>        Silence Clang static analyzer reports
>        cff-subset: Do not use garbage values
>        quartz: Silence some clang warnings
>        device: Flush on a finished device is a no-op
>        pattern: Complete the list of possible pattern errors
>        build: Respect CFLAGS and LIBS env settings
>        default-context: Do not allow restoring pushed gstates
>        clip: Fix copy of clip rectangles list
>        Fix make check
>        Update FSF address
>        Tighten error validation
>        surface: Set errors through _cairo_surface_set_status()
>        pattern: Do not provide type field in cairo_pattern_union_t
>        font-face: Cleanup backend-specific destruction code
>        Remove useless checks for NULL before freeing
>        hash: Improve double hashing
>        hash: Compare hash values before calling keys_equal
>        scaled-font: Remove _cairo_ft_scaled_font_get_unscaled_font()
>        scaled-font: Make unscaled font utility functions static
>        perf: Add hash table benchmark
>        hash: Improve handling of dead entries
>        hash: Code cleanup
>        clipper: Fix crashes
>        perf: Silence gcc warning
>        quartz: Fix compilation
>        win32: Fix compilation
>        time: Add cairo_time_t type
>        Introduce the cairo-missing library
>        perf: Reuse cairo-time
>        perf: Drop cairo_perf_ticks_t in favor of cairo_time_t
>        perf: Get rid of cairo_perf_ticks_per_second()
>        perf: Fix readme
>        script: Fix make check
>        surface: Fix make check
>        win32: Fix linking
>        perf: Fix win32 build
>        observer: Fix build on suncc
>        cairo-missing: Fix and cleanup ssize_t type definition
>        malloc: Fix build on suncc
>        wideint: Fix build on suncc
>        cairo-missing: Fix build
>        quartz: Fix build
>        Make cairo-missing a static library
>        quartz-font: Fix scaled font backend hooks
>        quartz: Fix the 32-bits build on MacOSX 10.7
>        rectangle: Fix warning
>        stroke: Fix typos
>        quartz-image: Fix build
>        vg: Fix build
>        cogl: Fix 'make check'
>        default-context: Fix 'make check'
>        xml: Fix backend structure
>        Silence some 'unused var' warnings
>        cairo-script: Don't ask for inlining of a cold-path function
>        boilerplate: Provide close callback after opening any2ppm output
>        Sort option flags
>        Improve the documentation of the flags
>        test: Re-enable pass-through
>        test: Remove dead code
>        test: Remove unused thread field
>        boilerplate: Remove unused thread id parameter
>        test: Reuse cairo_test_logv()
>        Use xstrdup instead of xmalloc when possible
>        Remove useless checks for NULL before freeing
>        boilerplate: Fix svg extension
>        test: Use cairo_test_list_t for the main test list
>        test: Do not open files in non-existing dirs
>        xcb: Fix typo
>        subsurface: Fix 'make check'
>        test: Use fork() on MacOS X
>        test: Add negative-stride-image
>        test: Add stride-12-image
>        quartz: Fix building with QUARTZ_DEBUG
>        quartz: Make glyph antialiasing consistent with quartz-font
>        quartz: Do not export private functions
>        raster-source: Do not return value in void function
>        pattern: Infinite color-only sources are always opaque
>        rectangle: Implement _cairo_rectangle_contains_rectangle()
>        observer: Return status when printing the observed data
>        observer: Silence gcc warning
>        win32: Fix Makefile.win32 build
>        device: Minor documentation fixes
>        build: Add 'clean' target to Makefile.win32 build system
>        test: Compile any2ppm on Windows
>        test: Fix any2ppm on MSVC
>        test: Create output directory even on Win32
>        build: Make 'all' the default target in the base Makefile.win32 build
>        path-fixed: Silence gcc warnings
>        stroke-style: Silence gcc warning
>        perf: Don't use a boolean value as integer
>        boilerplate: Use any2ppm from ANY2PPM env variable
> Behdad Esfahbod (4):
>        Don't err on non-invertible font matrix
>        Bug 34011 - headers-standalone target is incorrect for some non-gcc compilers
>        Don't cache _CAIRO_ENABLE results
>        [ft] Minor
> Benjamin Franzke (1):
>        egl: Fix eglMakeCurrent for egl surfaces
> Benjamin Otte (22):
>        AUTHORS: Use correct email for me
>        tests: Add line-width-tolerance
>        tests: Fix gcc complaint
>        test: Add line-width-overlap
>        xcb: Don't operate on a status variable that is never set
>        Assert that an error is valid when set
>        surface: Don't be nice to people setting internal error codes
>        device: Don't be nice to people setting internal error codes
>        region: Don't be nice to people setting internal error code
>        surface: Allow NOTHING_TO_DO as a valid error for now
>        test: Add missing format to any2ppm to silence gcc
>        surface: Actually return a value from _cairo_surface_set_error()
>        arc: Handle radius == 0 the same no matter the arc direction
>        Revert accidental checkin in last commit
>        Declare variables first so gcc shuts up
>        test: Add a test for 0-radius negative arcs
>        test: Fix gcc complaints about old style definitions
>        test: Fix copy/paste error
>        arc: Remove erroneous return statement
>        test: Fix gcc signedness warning
>        build: Use $GREP -e instead of plain grep -e
>        image: Don't crash on weird pixman formats
> Carlos Garcia Campos (1):
>        xlib: Check pixman format before trying to create an image surface for it
> Chris Wilson (533):
>        version: Post-snapshot bump
>        xcb: Only use clip rectangles for more than 1 pixel-aligned box
>        image: Limit span rendering to bounded extents
>        test: Attempt to capture out-of-bounds spans compositing bug
>        surface: Don't pass INT_UNSUPPORTED to _cairo_surface_set_error
>        configure: Test for funlockfile
>        xcb: Compile fix when using xcb-drm
>        trace: Fix hint_style/hint_metrics typo
>        image: Fix pixman format to RGB30 conversion
>        Silence the compiler for another couple of RGB30 switches
>        cairo-trace: Fix an obscure bug recording the fishtank
>        xlib,xcb: Force strict adherence to the Render specification when testing
>        boilerplate/xcb: Fix silly cut'n'paste errors in previous commit
>        test: Add unaligned-box
>        perf/micro: Add wave
>        perf
>        gl: Mark the use-once vertex buffers as DYNAMIC
>        boilerplate: Introduce create_similar hook
>        cairo-trace: Fix use of buf outside of scope
>        cairo-perf-diff-files: Don't print size/content for trace results
>        test: Add a Makefile.refs generator
>        xlib: Create an exact match for the image surface if possible using Xrender
>        png: Fix support of depth-30 images
>        test: Test a1 clipping semantics
>        ft-font: If the pattern is already resolved, use it immediately
>        scaled-font: Key the cache on the orignal font_face
>        Excise DRM-Xr
>        Excise xcb-drm
>        scaled-font: Fix assertions for original font-face vs font-face
>        image: Only discard the outer boxes
>        tests: Add clip-complex-shape
>        tests: Add paint-clip-fill
>        Implement cairo_backend_t
>        pattern: Add observer hooks
>        xcb: Remove more bits of drm integration
>        test: Always compile cairo-test-suite
>        test: Add random-clips to stress test clipping
>        gl: Compile fix, missing include cairo-pattern-private.h
>        tests,perf: Add a hatchings clip-test
>        test,perf: Another hatching!
>        ps: Add HAS_FT_FONT guards for type1
>        clip: Rudimentary support for clip-polygon extraction
>        win32: Fixup clip-mask fallback after API changes
>        clip-boxes: Intersection with 0 boxes means clip-all
>        xcb: Take advantage of clip-boxes
>        xcb: squash initial ChangePicture request for precision
>        xcb: Reduce one pass for masking, now just 2 passes required!
>        Fix proxying of ->context_create()
>        gstate: Handle NULL clip for in_clip()
>        xcb: Always send a clip region along with clipped glyphs
>        test: Expand partial-clip-text
>        test: Extend recording-surface-pattern to include a SOURCE operator test
>        test: Add a paint-with-alpha variant to test clip-boxes fast path
>        image: Fix partial clipping of text
>        image: replay the recording surface directly onto the target
>        image: Apply mask-opacity to clip boxes
>        tests: Missing ref images for new recording-surface test
>        tests: Another missing ref, this time clip-disjoint-hatching
>        region: Directly handle single rectangle creation in create_rectangles()
>        trace: Create a new opcode for recording surface
>        surface-wrapper: Fix use of uninitialised variable
>        clip: Embed a single box to avoid a common allocation
>        freed-pool: Enlarge the freed cache
>        scaled-font: Single glyph extent computations are worryingly frequent
>        composite: Perform a quick is-clipped for glyphs
>        recording: Only look for a fill-stroke operation if the target supports one
>        test: Exercise replaying basic painting paths
>        test: Add clip-intersect
>        image: Fix clip-intersect
>        xcb: set the right members of cairo_color_t for generating the opacity mask
>        recording: Move the glyph allocation into the wrapper after checking clip status
>        wrapper: Factor out the common clip handling
>        wrapper: Use the stack for small glyph allocations
>        wrapper: show-text-glyphs can now operate on constant array of glyphs
>        wrapper: Correct translation of clip for wrapper extents
>        clip: Skip processing of rectangle-intersection if it wholly subsumes the clip
>        clip: Short-cut the common condition of intersecting with a single box
>        recording: Combine the clip to the recording + target surface extents
>        xcb: Make composite_opacity_boxes slightly less silly
>        record: Store the untransformed operation extents along with the command
>        record: Check the operation against the target device extents.
>        wrapper: target to recording needs the inverse transform
>        rectangle: speed up the is-intersecting test.
>        API: map-to-image and create-similar-image
>        composite: Pass unbounded extents to initialisation
>        wrapper: translate the clip by the device transform
>        polygon: Don't skip clipped horizontal edges
>        polygon: Fix clipping of edges outside of their range
>        test: Add horizontal clip test
>        record: Use a bbtree to reduce is-visible checking overheads
>        record: And remember to offset the index of unbounded playback
>        record: We can only skip the clear so long as we know the destination is clear
>        record: Recording surfaces need an explicit clear
>        record: Assume recording surface targets are clear
>        clip: Fix cairo_clip_equal()
>        clipper: Detect a incremental change in the general clip-path
>        record: Only reduce fill+stroke if the clip doesn't change between the commands
>        fallback: Prevent recursion when combining with the clip
>        xcb: Short-circuit multiplying the alpha mask by 1.0
>        clip: Fix clip-equal to handle one or the other being NULL/all-clipped
>        clipper: Also need to guard against the incoming clip being NULL
>        pdf: Remove redundant clip regions
>        pdf: Don't remove the current clip if redundant
>        ps: Apply the clip reduction techniques from pdf
>        gl: Rectangular fast path
>        gl: Rectilinear fast path
>        xcb: Convert the box back to integer coordinates for uploading the image
>        trace: Emit the content type for image surfaces
>        image: extend support of direct replay for paginated surfaces
>        image: Free clip_surface after use
>        paginated: Remove an impossible test
>        test: Exercise extend modes with recording patterns
>        spans: fast-path common case of a single box.
>        tor: Use longjmp to throw an error whilst generating spans
>        image: Reduce compositing bounded boxes with a clip-mask to a polygon
>        clip: Mark __cairo_clip_all as private and do not export
>        tor: Sort the initial edge correctly
>        image: the boxes are already pixel-aligned so skip the extra rounding step
>        test: Extend PDF blend tests to include an opacity mask
>        tor: Suppress repeated spans
>        perf: Add a few more variations to dragon to exercise unaligned lines/spans
>        bo-rectangular: perform an incremental sort
>        perf: Save/restore gstate across runs
>        test: Add an test case to exercise overlapping caps between dash segments
>        test: Add a second ring to overlapping-dash-caps
>        bo-rectangular: handle in==out specifically for the single box case
>        perf: Add stroke/fill variants to world-map
>        test: Add world-map from the perf-suite
>        test: Add a couple of tests to exercise stroking of short tail segments
>        clipper: Don't emit an empty clip-path for no clip boxes
>        test: Add join-star, another stroker exercise
>        test: Add an loopy I-bar stroker test
>        test: Add joins-retrace
>        util: Add show-polygons
>        test: Add a1-sample
>        test: Add unclosed-strokes
>        perf: Add many-strokes
>        perf: add the usual special cases to many-strokes
>        tor: First perform a bucket sort before merge the sub-edges from the polygon
>        tor: Inline reverse insertion sort for handling intersections
>        build: Add a missing cairo-backend-private.h
>        perf: Add many-fills
>        fill: A horizontal/vertical line is also a degenerate fill box
>        bo-rectangular: Eliminate allocation for pqueue
>        bo-rectangular: Use a mergesort to speedup insertion
>        bo-rectangular: Correctly mergesort a doubly-linked list
>        tor: trivial changes
>        tor: Fix mergesort to handle doubly-linked list
>        perf: add many-curves
>        perf: add curve
>        tor: update is-vertical along with min-height
>        fixed: Allow the implicit close of the last fill path to complete a fill-box
>        trace: Pop the surface after write-to-png
>        script: Include an operator to replay a recording surface to a file
>        script: Compile fix
>        recording: do not reduce required clips
>        test: Add record-extend
>        util/show-polygon: Show end-points
>        util/show-traps: Cache the rendering of the traps+edges
>        script: Remove reference to image-surface-private
>        default-context: Tidy push-group
>        tor: Micro-optimise
>        bo: Perform an initial bucket sort on the start events
>        xlib: Set the clip_region for glyphs
>        test/line-width: Add a non-antialiased variant
>        perf/micro: Test wide vs hairline strokes
>        perf/micro: diagonal lines
>        bo-rect: Micro-optimisation
>        image: move surface definition to new header for subclassing
>        skia: Update to use cairo_backend_t interface
>        Fix pollution from skia commit
>        surface-wrapper: Initialise clip to NULL
>        bo-rect: One step too far...
>        boilerplate: improve fidelity of surface extraction
>        recording-surface: Don't store the transient error when returning the path
>        surface: propagate internal statuses
>        pdf: Propagate NOTHING_TO_DO
>        surface: Don't modify operator
>        recording: replay_all is meant to mean REPLAY&&  ALL!
>        pdf: Compute fill-stroke extents first before trying to use it to set the clip
>        pdf: composite-rectangles now require freeing after use
>        Only reduce the clip if it is not in active use for the operation
>        test/line-width: Refactor and tidy
>        test: Add a couple of variants to line-width-overlap
>        stroke: Rely on the tessellator to remove self-intersections
>        Add missing 'cairo-image-surface-private.h'
>        test/xlib-expose-event: Be kinder to recording surfaces
>        script: Emit sequences of boxes to as 'rectangle' for clarity
>        snapshot: Defer acquisition
>        script: Support unbounded native recording surfaces
>        recording: remove the duplicate 'content' field
>        recording: break self-copy loop
>        subsurface: call the high-level cairo_surface_flush
>        subsurface+recording: handle recursion
>        image: peek through a snapshot to the recording surface behind
>        paginated: unwrap subsurfaces during context creation
>        analysis: prevent recursion whilst analysing recording patterns
>        ps: unwrap recording surface snapshots
>        snapshot: Assert that we do not generate a snapshot clone
>        paginated: Use the backend->snapshot
>        wrapper: Use the backend->snapshot function
>        pdf: If the recording surface is unbounded, limit the pattern to the ink extents
>        script: leave the tail of the RGB24 data unmolested
>        script: take advantage of the polymorphism of the interpreter
>        script: Missed break for creating unbounded recording surfaces.
>        xlib: Mark surfaces as finished when the Display is finished/destroyed/closed.
>        xlib: Move the Display pointer nullify into destroy from finish
>        record: Offset the clip by the replay transformation as well
>        surface: Rearrange nothing_to_do? to catch OVER + clear source
>        script: Hook image caching into the snapshot mechanism
>        gstate: Copy clusters for an untransformed unbounded surface
>        Introduce cairo_surface_observer_t for performance analysis
>        script: enable by default
>        tee: compile fix for migration of _cairo_is_recording_surface()
>        xml: Include 'cairo-image-surface-private.h'
>        Introduce cairo_mime_surface_t
>        check: make check-headers happy
>        recording: Defend against bad user-input
>        stroke: move normal stroker to new file
>        stroke: Convert the outlines into contour and then into a polygon
>        observer: report number of solid patterns first
>        observer: print operator frequencies
>        observer: print stroke caps/joins frequencies
>        snapshot: restore the order of detach vs callback
>        observer: print path antialias modes
>        obverser: only print out the active patterns
>        observer: only print out the non-zero path types
>        observer: only print out the non-zero clip types
>        observer: further classify general clips
>        observer: put a comma between array items when printing
>        obsever: include the operation timings
>        perf: Cleanup target after each run
>        observer: record all operations and their timings
>        xlib: Fix typo in snapshotting.
>        recording: optionally disable optimise away clears
>        observer: copy glyphs around call into backend
>        observer: bypass surface mid-layer and call into recording surface directly
>        test: Add checkerboard
>        test: Add big-little-box
>        test: Add record-mesh
>        test: Extend record-extend to exercise similar surface sources
>        test: Extend rotate-image-surface-paint
>        test: Add shape-general-concave
>        observer: correct classification of aligned paths
>        observe: Provide the sum of the elapsed time of the individual operations
>        test: Add clip-mixed-antialias
>        test: Add big-little-triangle
>        test: Add big-empty-box
>        test: Add big-empty-triangle
>        perf: Print a summary of each operation for a trace (using '-s')
>        api: Extend cairo_antialias_t to include performace/quality hints
>        observer: Add missing return codes
>        perf/Makefile.am: Add missing '\' line continuation
>        Introduce a new compositor architecture
>        test-surfaces: compilation fixes
>        perf/micro: Add measurement of setting the pixel directly using pixman
>        test: Update ref images
>        mono-scan-converter: Decrement height as we skip straight edges
>        perf/stats: Avoid overflow when summing time-squared
>        traps: Send unaligned boxes as trapezoids
>        traps: Skip compositing an empty bounded regions
>        image: Adapt to updated pixman lerp operators
>        test: Add the PS tiger
>        perf/micro: Add the PS tiger as a measure for the antialias hints
>        mono: Amalgamate adjacent spans
>        skia: fix compilation
>        image: fix compilation of spans with LERP
>        test: Fix ref images for a1-clip-*
>        test: Add ref image for random-clip
>        perf: Add an a1-pixel variant
>        test: Use the trapezoid reference image for random-clip
>        test: Complete set of baseline reference images
>        ps: Set transparency for stencil_masks
>        test: Move all the reference images to their own directory
>        test: Write the individual test logs to output/
>        ps: Fix return value for mask_supported and ps2
>        ps: Check earlier for a supported mask
>        test: Mark some more raster-only tests
>        composite: Reduce an unaligned clip
>        script: Tag the similar surface rather than snapshot
>        test: Fix reference image for unbounded operator
>        build: Add missing cairo-surface-backend-private.h to sources
>        build: Another missing private.h
>        check: Make the newly added privates pass make check
>        image: Invert recording matrix before replay
>        test: refresh clip-operator ref.png
>        test: Record the failure of recent freetype libraries for type1 vertical layout
>        test: Update refernece dir for create-from-png-stream
>        test: Update couple of refs for the improved mono rasteriser
>        test: Use the test-traps as the reference images for xlib/xcb
>        test: Replace xlib reference images with the traps references
>        fdr,tee: Reorder master/slave invocation to capture death-upon-signals
>        test: Hack cairo-test-trace to write at trace for all contexts
>        test: Add bug-bo-ricotz
>        bentley-ottman: End subsumed colinear traps
>        test: Update reference images for test-base
>        base: Make sure we have fuzzy clip boxes!
>        path: Skip calls to zero-length memcpy
>        test: Add a1-fill
>        test: Refresh reference images for spans/traps
>        image: Use A8 for creating a mask from a recording surface
>        image: Use the recording surface content for the recording source
>        wrapper: intersect with target extents
>        image: intersect the source extents with the replay extents for EXTEND_NONE
>        test: Record the current status of radial-outer-focus as xfail
>        composite-rectangle: Add a helper to refine source extents
>        pdf: Use the helper functions to update the composite rectangles
>        os2: Blindly update os2 to the new fallback compositor
>        test: Make cairo_test_suite depend upon the any2ppm exectuable on all platforms
>        traps: use the customised _cairo_clip_get_surface
>        xcb: track fallback damage
>        wrapper: transform the clip into device space
>        image: clip the replay to the sample extents in device space
>        image: don't offset by device transform for replay onto source surface
>        test/trace: Hack to dump out per-context images and traces
>        test/create-from-png: Update to point to new reference/ images
>        test: refresh text-pattern reference
>        test: Exercise scaling from an atlas through a subsurface
>        test: Redefine success for miter-precision
>        spans: Fix empty polygon unbounded fixup
>        test: Refresh tighten-bounds reference image
>        test: Refresh rotated-clip reference
>        test: Mark the record-select-font-face as XFAIL
>        boilerplate: Skip testing of null surfaces
>        win32: Compile, but broken
>        script: compile without
>        polygon-intersect: Remove surplus edge direction
>        test/random-clips: Paint clip to highlight the issues
>        test: Refresh reference images
>        trace: Emit an stack operation to create a pattern from an undefined surface
>        test: Add line-width-large-overlap
>        stroke: Use the tessellator to resolve overlapping strokes
>        image: Enable use of LERP_SRC for masked source composition
>        fill: Fix unantialiased rectilinear-fill-to-boxes
>        test: Add clip-rectilinear
>        gl: Declare coverage input for fragment shaders.
>        gl: Basic fixes to get cairo-gl running again
>        gl: Need to increment reference count when copying operands
>        botor: Remove a couple of inlines to cleanup -Winline
>        perf
>        debug-traps
>        gl: Include use-coverage in shader hash
>        gl: Spans start after the mask, not coincident!
>        gl: Remove unused alpha argument from operand_emit
>        xcb: Re-enable glyph rendering
>        observer: Fix typo and handle the condition of no script device
>        mono: Silence valgrind by placing a sentinel value in the sorted buckets
>        spans: Propagate internal status when retrieving the clip surface
>        test: Add clip-polygons
>        surface-clipper: Fix path leak due to typo
>        clip: Free the freed clip pool on reset
>        clip: Replace the original clip when transforming
>        recording-surface: Free the contents of the command array when clearing
>        recording-surface: Optimize away anything below an opaque fill
>        spans: Refresh polygon limits after trimming the composite extents
>        gstate: Prevent leak of old clip when creating a group with translation
>        recording-surface: Initialize optimize-clears before use in snapshotting
>        ft: Add missing break to enable BGR subpixel rendering
>        gl/msaa: Markup the new symbols as private for PLT hiding
>        xlib: Reduce the composite traps operator for when the dest is clear
>        xlib-xp
>        Merge branch 'master' of git://cairographics.org/git/cairo
>        Merge branch 'master' of git://cairographics.org/git/cairo
>        gl: Initialize spans on the context
>        Revert "xcb: Fix xcb-huge-image-shm"
>        xcb: Handle SHM exhaustion gracefully
>        perf: Compile fix, add the index to cairo_perf_report_load()
>        polygon-intersection: Finish any edges upon intersection
>        image: Only unwrap a subsurface if the sample is fully contained
>        subsurface: Simplify acquire_source_image
>        image: Remove dubious "optimisations" for acquired source images
>        subsurface: And remove the dead code for releasing complex source
>        xlib: ADD only reduces to SOURCE for alpha-only targets
>        xlib: Handle subsurfaces correctly
>        subsurface: Support caching for cloned subsurfaces
>        surface: Bump reference count around finish
>        xlib: Cache the subsurface Picture
>        xlib: Eliminate redundant copies of subsurfaces
>        gl: Update glyphs to use cairo_gl_source_t
>        gl: Correctly offset sub-image uploads
>        gl: Fallback for complex subsurfaces
>        gl: Avoid cleaning up an uninitialised composite upon UNSUPPORTED
>        gl: Embed a operand into the surface for pass-through
>        gl: Prevent the traps compositor from exploding
>        Convert cairo_mime_surface_t to cairo_raster_source_pattern_t
>        script: Prefer polymorphorism for mesh path construction
>        trace: Remove the warning about the unstable format
>        gl: Temporarily clone a subsurface as required
>        gl: And enable subsurface caching of the clones
>        subsurface: Replace any existing snapshot
>        gl: Even repeating subsurfaces need to be cloned
>        xcb: Silence a compiler warning for mixing status and internal status enums
>        doc: Add similar-image, map-to-image, unmap-image
>        doc: Add sections for cairo-script
>        script: Add documentation
>        doc: Add new antialias symbols
>        doc: add CAIRO_DEVICE_TYPE_COGL
>        doc: Drop the gtk-doc markup from _cairo_radial_pattern_focus_is_inside
>        gl: Make the backend struct static
>        gl: Check against user-provided invalid sizes
>        gl: Propagate surface texture to embedded operand
>        gl: Use the embedded operand to allow passing sources around
>        gl: Embed the operand rather than a pattern into the glyph cache
>        gl: Decouple glyphs on shutdown from the scaled font caches
>        gl: Substitute the white source for the default pattern
>        gl: Propagate clip region
>        gl: Set the device offset on map-to-image
>        fallback: fix the offset for painting
>        gl: Unbreak the glyph cache
>        gl: Re-enable the CLEAR optimisation
>        polygon: Assert that we add edges that are wholly contained by the clip
>        polygon: Tweak the y-coordinates of the edge so that it is inside the clip
>        test: reference ref images after tweaking polygon clipping
>        test: Exercise copy/filling unsorted rectangles
>        gl: Defer stencil allocation until use
>        gl: Prevent leak of the white source used with glyph masks
>        doc: Add documentation for cairo_raster_source_pattern
>        gl: Decouple the glyph upon eviction
>        gl: Track surface references through operands
>        gl: Set the destination for swap buffers, required by EGL at least
>        gl: Disable the msaa compositor by default (for release testing)
>        perf/chart: Tweak labels on right not to fall off the edge
>        perf/chart: Show values next to the column if too small to fit inside
>        perf/chart: Make the columns transculent so that the label behind is visible
>        test: Add arc-direction
>        xlib: Do not upload inplace if the image does not match the surface format
>        xcb: Add dimension guards to create-similar-image
>        subsurface: Add guards for creating similar surface
>        xlib: Fix typo in 5045155de6, lack of closing ';'
>        xlib: Improve choice of bits-per-pixel for depth
>        xlib: Set IncludeInferiors when using CopyArea
>        xlib: Set IncludeInferiors when acquiring the source image
>        xlib: Only reduce a readback of an uninitialised source for pixmaps
>        gl: Just flush the context upon operand destroy
>        Replace the ad-hoc surface unwrappers with a function pointer
>        test: Referesh traps (xlib) reference images for font updates
>        directfb: Discard long broken code and return to basics
>        polygon: Extend intersection edges to cover entire range
>        tor: Restore the 256x15 sampling of the original rasteriser
>        clip: Check whether an extents only clip contains the box
>        clip: Apply the partial boxes for clip_combine_with_surface
>        spans: Pass unbounded operations to the spans compositors
>        traps: composite_boxes() is not a mask constructor
>        polygon-intersection: The edge direction is immaterial
>        gl: Transfer ownership of trapezoid mask to operand
>        analysis: replace open-coded _cairo_box_add_box()
>        quartz: Add missing source hook
>        Add preliminary damage tracking
>        win32: Move to separate directoy
>        win32: Rebase on the new compositor infrastructure
>        win32: Fix lifetime tracking of create_similar_image()
>        win32: Cast the surface to an image-surface to find its parent (compile fix)
>        test: Refresh reference images for antialiasing fix
>        image: Add the get-font-options callback again.
>        test/README: Update sha1sums for fonts used
>        test: Update base reference images for new reference font
>        test: Update traps for changes in reference font
>        boilerplate/xlib: The xlib-fallback should be visually identical to image
>        test/xlib: Remove stale reference images
>        xlib: Replace obsolete disable-xrender with shiny new device debug interface
>        test: Add a set of reference images for the mask-based compositor
>        boilerplate/xlib: Exercise the mask based compositor for xrender version 0.0
>        xlib: Hook up copy_boxes for the mask compositor
>        mask-compositor: Acquire the target surface when creating the composite mask
>        traps-compositor: Reduce a complex clip generation to an ADD operator
>        xlib: Trim the intermediate upload image to match the upload extents
>        xlib: Correct the device-offset applied to the map-to-image result
>        xlib: Make the core compositor actually paint
>        xlib: Trim the ximage to match the trimming of the intermediate
>        traps-compositor: add a missing release()
>        script: Update mesh pattern for earlier change of csi operators
>        image: Tidy reduction to EXTEND_NONE for replay surfaces
>        surface-wrapper: Transform the clip by the device-transform correctly
>        xlib-xcb: Wrap cairo_xlib_device_debug_cap_xrender_version
>        xlib: Tidy conversion of xrender format to cairo_content_t
>        xlib: Handle window-to-window copies by avoiding the use of a clip region
>        {mask,traps}-composite: Restore unsetting of the is-clear flags for the mask
>        image: Handle recursive drawing of recording surface
>        debug: Add some TRACE statements for recording surfaces and snapshots
>        recording: Optimize a copy of one recording surface onto another
>        subsurface: Fix typo in snapshot creation
>        test: Add a couple of reference images for the recording surface
>        recording: Remove superfluous reset
>        test: a1-line-width is a test of rasterisation, so skip for the vector backends
>        pdf,ps: The device offset only needs to taken into account for raster patterns
>        test: Verify scaled replay of a recording surface
>        surface-wrapper: Apply replay transforms to scaled font
>        test/record: Fix typo in record*-text-transform
>        test: Set record2x-fill-alpha as an xfail
>        scaled-font: Key the cache on the original font face
>        scaled-font: Refactor the hash computation
>        test: Add record90 to exercise replaying a rotated recording surface
>        scaled-font: Assign a temporary hash value for the placeholder
>        test: Regenerate reference images for vanilla record tests
>        surface-wrapper: Apply the scaled-font ctm and non-default font-options
>        spans+image: Fix clipping with polygons and spans
>        gl: Various fixes for glyphs
>        ft: Remove unused HINT_METRICS from ft_font.extra-flags
>        ft: Export FreeType synthesis options
>        spans,traps-compositor: Check for all-clipped after intersecting clip with boxes
>        test/record1414: Pixel align the clip for replay
>        base: Remove the double application of the clip boxes
>        image: Correct origin of unbound recording surface source
>        spans: Pass antialiasing hint down to the backends
>        image: Add a fast path for solid-filled polygons
>        spans: Retrim extents to clipped polygon
>        image: Perform the general composite operation inplace for mono rasterisation
>        image: Add a fast-path for mono-rasterised blits
>        spans: Reduce composite_aligned_boxes with over to source for opaque patterns
>        spans,image,gl: Add fast-path for simple copies
>        traps: Use the mono-scan-converter to reduce the number of traps
>        traps: First attempt to convert curvy unantialiased polygon to scanlines
>        spans: Handle fallbacks from upload-boxes by reverting to the normal composite
>        image: Perform the simple solid-fill spans inplace
>        image: Add a simple inplace blitter for spans
>        image: Try performing span composition a row at a time
>        surface-wrapper: Only apply the wrapped transform to the scaled-font
>        image: Add unbounded support to the mono rasteriser fast-paths
>        clip: Use the boxes-intersection routine for computing the clip polygon
>        traps: Avoid double application of unaligned clip boxes
>        traps: The CompositeTrapezoids requires the unbounded fixup for clipping
>        xlib: Apply the glyph offset
>        surface: Relax assertion about not rendering to a snapshot
>        test: Refresh the fallback-resolution reference images
>        hash: Keep a simple lut in front of the main hash
>        bentley-ottmann: Skip intersection check if the bounds do not overlap
>        bentley-ottmann: Sort by edge bounding boxes before computing x
>        mono-scan-converter: Use edge->is_vertical flag
>        stroke: Do not initialise the pen if will not use it
>        skia: compile fix
>        test: Restore 'release-verify-sane-tests' makefile target
>        version: Bump for snapshot 1.11.4
>        version: Post-release version bump to 1.11.5
> Chuanbo Weng (1):
>        gl: fix bug in _cairo_gl_surface_embedded_operand_init()
> Dagobert Michelsen (1):
>        Use detected EGREP instead of generic grep
> Dave Yeo (3):
>        os2-surface needs to include cairo-image-surface-private.h
>        Fix the number of parameters for cairo_os2_surface_set_size()
>        On OS/2, WIFSIGNALED and WTERMSIG are prototyped in sys/wait.h
> Ehsan Akhgari (1):
>        Avoid defining inline when compling C++ on MSVC.
> Eric Anholt (4):
>        gl: Don't reset the FBO draw/readbuffers every time we bind the FBO.
>        gl: Add a first bit of general documentation on cairo-gl usage.
>        gl: Take advantage of GLX_MESA_multithread_makecurrent to avoid unbinding.
>        gl: Bind samplers just once at program compile time.
> Haithem Rahmani (1):
>        boilerplate: Use correct flag type in DFBWindowDescription
> Henry (Yu) Song (1):
>        gl: initialize temporary surface with size same as subsurface
> Jeff Muizelaar (5):
>        Add UINT32_MAX define.
>        Add an implementation of fmax for use with MSVC
>        Remove quartz-image extents.
>        win32: Explain the purpose of the surface extents
>        Avoid implicit conversion to double when rounding
> Jesse Barnes (1):
>        Expose 30bpp/10bpc support: CAIRO_FORMAT_RGB30
> Jesse van den Kieboom (1):
>        quartz: Rename declaration of _cairo_quartz_create_cgimage
> Keith Packard (4):
>        cairo-xcb: gtk-doc doesn't like _ in parameter names
>        Create XCB documentation.
>        Add cairo_xcb_surface_set_drawable
>        Clip rectangles are not necessarily YSorted
> Krzysztof Kosinski (1):
>        test: Add radial-outer-focus
> Krzysztof Kosiński (1):
>        test: Add test case from bug-40410
> M Joonas Pihlaja (1):
>        image: Avoid reusing pixman images for threadsafety.
> Martin Robinson (14):
>        gl/msaa: Introduce an MSAA compositor for OpenGL
>        gl/msaa: Implement basic solid color fill
>        gl/msaa: Add clipping support
>        gl/msaa: Support for solid color strokes.
>        gl/msaa: Fix glScissor bounds
>        gl/msaa: Fix scissor bounds
>        gl: Fix offset for non-texture surface patterns
>        gl: Properly prepare the source when compositing glyphs with a mask.
>        gl/msaa: If stenciling the clip fails, reset the color mask.
>        gl/msaa: Ignore the antialiasing property of the clip.
>        gl/msaa: Scissor simple rectangular clips
>        gl: Fix the offset of non-texture source surfaces
>        gl: Fix gl-source-surface test
>        gl/msaa: Do not attach a depth-stencil attachment to the default framebuffer
> Matt Peterson (1):
>        gl: Use cairo_rectangle_int_t
> Murray Cumming (2):
>        cairo: Remove trailing comma from cairo_format_t
>        cairo_surface_observer_mode_t: Remove trailing comma.
> Nis Martensen (24):
>        Fix typos
>        Mark the new API as such
>        doc: Fix pattern name mismatch
>        doc: fix typo
>        doc: add missing colons to since tags
>        doc: typo
>        doc: add index of new symbols in 1.12
>        doc: add colon to .._padded_image_from_image
>        doc: add missing _face to function names
>        doc: do not confuse gtk-doc by @ and double *
>        doc: Add colon to Returns statement
>        doc: script_mode_t needs at least one line of text
>        doc: fix a few typos found by codespell
>        doc: preserve whitespace by using docbook screen tag
>        doc: symbol names in header and comment must match
>        doc: ensure "compatibility" spelling
>        doc: move value descriptions up
>        doc: avoid confusing gtk-doc by double asterisks
>        doc: describe cairo_recording_surface_get_extents
>        doc/ft-font: add reference to cairo_ft_synthesize_t
>        doc/ps-surface: minor improvements
>        doc/xlib/xcb: document get/set_precision API
>        doc/xcb: document cairo_xcb_device_get_connection
>        doc: fix broken link
> Oleg Romashin (1):
>        qt: Fix compilation, also minor fix for building against Qt5
> Paulo Zanoni (3):
>        test: fix append_argv()
>        test/README: add missing "S"
>        cairo-test-runner: don't leak argv
> Pekka Paalanen (1):
>        egl: do not destroy dummy_surface if it does not exist
> Pino Toscano (1):
>        LD_PRELOAD is supported on Hurd
> Rob Bradford (1):
>        build: Specify the reference images as a directory for EXTRA_DIST
> Robert Bragg (1):
>        backends: Adds a new Cogl based backend
> Seongwon Cho (1):
>        mono-scan-convertor: Include space for the closing span
> Søren Sandmann Pedersen (1):
>        Include pixman.h in cairo-xlib-private.h
> Taekyun Kim (4):
>        tessellator: Fix boxes tessellator to handle num_boxes<= 1 correctly
>        tessellator: Fixed to produce an output box with x1<= x2 for single box input
>        Fix intersect_with_boxes() to produce tight clip extents
>        image: Maximum number of spans can be upto (extents->width + 1)
> Uli Schlachter (131):
>        xcb: Fix a BadPicture when clearing a surface
>        xcb: Automatically enable the backend if the libs are available
>        xcb: Document all public functions
>        User fonts: Make it clear what should NOT be freed
>        scaled font: Fix the docs for cairo_scaled_font_get_font_face
>        xcb: Limit the amount of SHM used
>        xcb: Remove a duplicate static function
>        xcb: Move the allocation of a shm surface into its own function
>        xcb: Use a normal image surface if SHM fails
>        XCB: Fix build with xcb-shm disabled
>        ps: Update xlib-surface-source ref images
>        xcb-surface-source: Add image16 and ps ref images
>        Finish devices after their finish callback returns
>        xcb: Fix a crash when finishing a device
>        xcb: Remove the surface pattern cache
>        Fix errors from src/check-plt.sh
>        arc-looping-dash: Add a missing cairo_restore ()
>        xcb: Remove some unused code
>        xcb: Fix for PDF operators with RENDER 0.11
>        xcb boilerplate: Handle device offsets correctly
>        xcb: Remove _cairo_xcb_picture_copy
>        xcb: Never return NULL from create_similar
>        xcb-drm: Fix undeclared variable error
>        Add unaligned boxes to "operator-source" test
>        image: Don't use the fast path if it's wrong
>        test/Makefile.am: Don't reference non-existant files
>        Revert "xcb: Never return NULL from create_similar"
>        XCB: Implement PDF blend operators natively
>        xcb: Fallback to image surface for create similar
>        xcb,xlib,surface: Check for too small sizes
>        xcb: Don't hardcode the RENDER version number
>        xcb: Prettify some code
>        xcb: Initialize the new precision fields
>        xcb: Use defines instead of magic numbers
>        xlib-xcb: Use slim_hidden_proto correctly
>        xlib-xcb: Fix api-special-cases test
>        xlib-xcb: Verify we really have an xcb surface
>        xlib-xcb: Fix cairo_surface_flush()
>        xcb: Remove an unused function argument
>        xlib-xcb: Fix some use-after-free
>        xlib-xcb: Don't call directly into the xcb backend
>        Xlib: Fix boilerplate to work with xlib-xcb
>        xlib-xcb: Register a XCloseDisplay hook
>        xcb: Move cairo_xcb_picture_t to cairo-xcb-private.h
>        xcb: Track cairo_xcb_picture_t surfaces
>        Add a test case that asserts on xcb
>        xcb: Don't use xcb surfaces as snapshots
>        Handle CAIRO_STATUS_DEVICE_FINISHED in switches
>        cairo-trace: Fix mark-dirty with xcb backend
>        xlib-xcb: Fix an "extension leak"
>        cairo-xcb: Drop some unused definitions/file
>        xcb: Fix _set_clip_region for many rectangles
>        xcb: Fix a GC leak when a screen is finished
>        xlib-xcb: Fix 'incompatible pointer type' warnings
>        XCB: Store the flags per-connection only
>        xcb: Make it possible to undo _cairo_xcb_device_debug_cap_*
>        Xlib,xcb: Make *_debug_[sg]et_precision() more robust
>        xcb: Parse $CAIRO_DEBUG just like cairo-xlib does
>        xcb: Fix a NULL dereference
>        xcb: Handle deferred_clear in _get_image
>        xcb: Unset the deferred_clear flag on fallback
>        xcb: Work around wrong extent computation in the X server
>        xlib: Fix a typo
>        xcb: Check that the extents are inside our surface
>        Make _cairo_xcb_surface_get_extents no_warn
>        xcb: Assert that pixmap sizes are positive
>        xcb: Error on 0x0 source surfaces
>        xcb: Handle deferred clear in map_to_image
>        xcb: _get_image only works without fallback
>        xcb: Handle fallback in map_to_image
>        xcb: Handle deferred clear in _upload_image_inplace
>        surface_unmap_image: Fix fallback
>        xlib-xcb: Implement the new backend functions
>        perf: Also build the code in perf/micro
>        Clarify the API docs for the newest functions
>        cairo.h: Document CAIRO_DEVICE_TYPE_INVALID
>        xlib-xcb: Fix a double free in surface_unmap
>        map_to_image: Verify the given extents
>        map-to-image: Handle non-32bpp formats
>        xcb: Fallback to image if allocating SHM fails
>        xcb: Merge two functions for creating shm images
>        test: Add a test that maps a huge surface
>        xcb: Handle SHM exhaustion via falling back
>        xcb-shm: Fix a logic error while allocating mem
>        test: Add tighten-bounds
>        xcb: Add an assert for the tighten-bounds test
>        xcb: Fix fallback for *_shm_put_image
>        xcb: Skip drawing if bounded extents are empty
>        xcb: Fix fixup_unbounded_with_mask
>        xcb: Steal from the pending list for GetImage
>        cairo_clip_path_t: Remove extents
>        xcb: Remove some dead code
>        xlib-xcb: Make this compile again
>        Tee: compile fix for recent compositor API
>        Revert "xcb: Error on 0x0 source surfaces"
>        xcb: Improve error cases in _clip_and_composite_combine
>        cairo-xcb: Require libxcb 1.6
>        boilerplate-xcb: Print sequence numbers
>        path: Fix a minor oversight in 1ce5d4707cf26061
>        _cairo_clip_get_surface(): Don't lose errors
>        xcb: Stop using _cairo_clip_get_surface()
>        xcb: Use a mask in _composite_boxes() when needed
>        xcb: Remove an unused variable
>        xcb: Don't try to fallback more than once
>        xcb: Honor clips for defer clear
>        xcb: Fix a "incompatible pointer" compiler warning
>        xcb: Fix device offsets with unmap_image()
>        test: Add mime-surface-api
>        Make the new mime-surface-api succeed
>        xlib: Fix compilation with --disable-xlib-xrender
>        create-from-png*: Test mark_dirty with mime data
>        flush: Detach mime data
>        test: Add clip-double-free
>        clip: Fix clip-double-free
>        clip_intersect_boxes: Fix memleak
>        test: Add text-antialias-subpixel-{,v}{bgr,rgb}
>        xcb: Fix some invalid casts
>        xcb: Fix xcb-huge-image-shm
>        xcb: Fixup some internal state in set_{drawable,size}
>        xlib-xcb: Implement surface_set_drawable
>        xcb: Fix invalid casts from cairo_content_t to cairo_format_t
>        xcb: Silence a compiler warning for mixing type and internal type enums
>        xcb: Silence compiler warnings about ignored return values
>        xcb: Move the surface picture setup into its own function
>        xcb: Add a special case for recording surface
>        xcb: Use int instead of uint16_t for rowstride
>        xlib-xcb: Make this compile again
>        Wrapper: Don't translate clips extents' origin
>        Fix docs for cairo_xlib_device_debug_cap_xrender_version
>        xlib-xcb: Fix make check
> Zhigang Gong (1):
>        mono: Always bias initial edge advancement
> --
> cairo mailing list
> cairo at cairographics.org
> http://lists.cairographics.org/mailman/listinfo/cairo

More information about the cairo mailing list