AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
82dec0e16a
skia2/tools/pathops_sorter.htm

1439 lines
54 KiB
HTML
Raw Normal View History

Enabling the canvas bit to turn the clip stack into a flat replace exposed around 100 failures when testing the 800K skp set generated from the top 1M web sites. This fixes all but one of those failures. Major changes include: - Replace angle indices with angle pointers. This was motivated by the need to add angles later but not renumber existing angles. - Aggressive segment chase. When the winding is known on a segment, more aggressively passing that winding to adjacent segments allows fragmented data sets to succeed. - Line segments with ends nearly the same are treated as coincident first. - Transfer partial coincidence by observing that if segment A is partially coincident to B and C then B and C may be partially coincident. TBR=reed Author: caryclark@google.com Review URL: https://codereview.chromium.org/272153002
2014-06-17 12:15:38 +00:00
<!DOCTYPE html>
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
<html lang="en" xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8" />
<title></title>
<div style="height:0">
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect1">
{{{1.80943513, 3.0778243500000002}, {1.66686702, 2.1680693600000001}, {1.68301272, 0}, {3, 0}}} id=1
{{{0, 1}, {0, 2}, {0.75, 2.25}, {1.75, 2.25}}} id=2
{{{1.75, 2.25}, {2.75, 2.25}, {4, 2}, {5, 2}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect2">
{{{1.80943513, 3.0778243500000002}, {1.66686702, 2.1680693600000001}, {1.68301272, 0}, {3, 0}}} id=1
{{{0, 1}, {0, 2}, {0.75, 2.25}, {1.75, 2.25}}} id=2
{{{1.75, 2.25}, {2.2500000000000018, 2.25}, {2.8124999999999991, 2.1875}, {3.375, 2.125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect3">
{{{1.80943513, 3.0778243500000002}, {1.738151075, 2.6229468550000004}, {1.7065454725, 1.8534907675000001}, {1.85738429375, 1.19775405375}}} id=1
{{{0, 1}, {0, 2}, {0.75, 2.25}, {1.75, 2.25}}} id=2
{{{1.75, 2.25}, {2.2500000000000018, 2.25}, {2.8124999999999991, 2.1875}, {3.375, 2.125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect4">
{{{1.80943513, 3.0778243500000002}, {1.7737931025, 2.8503856025000003}, {1.7480706881249999, 2.5443022068750003}, {1.7501136332812499, 2.2131114089062502}}} id=1
{{{0, 1}, {0, 2}, {0.75, 2.25}, {1.75, 2.25}}} id=2
{{{1.75, 2.25}, {2.2500000000000018, 2.25}, {2.8124999999999991, 2.1875}, {3.375, 2.125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect5">
{{{1.80943513, 3.0778243500000002}, {1.7737931025, 2.8503856025000003}, {1.7480706881249999, 2.5443022068750003}, {1.7501136332812499, 2.2131114089062502}}} id=1
{{{0.5, 2}, {0.81249999999999956, 2.1874999999999987}, {1.2500000000000009, 2.2500000000000013}, {1.75, 2.25}}} id=8
{{{1.75, 2.25}, {2.2500000000000018, 2.25}, {2.8124999999999991, 2.1875}, {3.375, 2.125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect6">
{{{1.80943513, 3.0778243500000002}, {1.7737931025, 2.8503856025000003}, {1.7480706881249999, 2.5443022068750003}, {1.7501136332812499, 2.2131114089062502}}} id=1
{{{0.5, 2}, {0.81249999999999956, 2.1874999999999987}, {1.2500000000000009, 2.2500000000000013}, {1.75, 2.25}}} id=8
{{{1.75, 2.25}, {1.9999999999999976, 2.25}, {2.2656250000000022, 2.234375}, {2.5390625, 2.2109375}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect7">
{{{1.80943513, 3.0778243500000002}, {1.7737931025, 2.8503856025000003}, {1.7480706881249999, 2.5443022068750003}, {1.7501136332812499, 2.2131114089062502}}} id=1
{{{1.0546875, 2.1953125}, {1.2656250000000009, 2.2343750000000009}, {1.4999999999999996, 2.2499999999999996}, {1.75, 2.25}}} id=12
{{{1.75, 2.25}, {1.9999999999999976, 2.25}, {2.2656250000000022, 2.234375}, {2.5390625, 2.2109375}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect8">
{{{1.7656425168945311, 2.6843748983789064}, {1.7550120280078125, 2.5380253562890633}, {1.7490921607031253, 2.3787068078906248}, {1.7501136332812499, 2.2131114089062502}}} id=7
{{{1.0546875, 2.1953125}, {1.2656250000000009, 2.2343750000000009}, {1.4999999999999996, 2.2499999999999996}, {1.75, 2.25}}} id=12
{{{1.75, 2.25}, {1.9999999999999976, 2.25}, {2.2656250000000022, 2.234375}, {2.5390625, 2.2109375}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect9">
{{{1.7656425168945311, 2.6843748983789064}, {1.7550120280078125, 2.5380253562890633}, {1.7490921607031253, 2.3787068078906248}, {1.7501136332812499, 2.2131114089062502}}} id=7
{{{1.0546875, 2.1953125}, {1.2656250000000009, 2.2343750000000009}, {1.4999999999999996, 2.2499999999999996}, {1.75, 2.25}}} id=12
{{{1.75, 2.25}, {1.8750000000000016, 2.2499999999999991}, {2.0039062499999982, 2.2460937500000004}, {2.1357421875, 2.2392578125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect10">
{{{1.7656425168945311, 2.6843748983789064}, {1.7550120280078125, 2.5380253562890633}, {1.7490921607031253, 2.3787068078906248}, {1.7501136332812499, 2.2131114089062502}}} id=7
{{{1.3876953125, 2.2373046875}, {1.50390625, 2.24609375}, {1.625, 2.25}, {1.75, 2.25}}} id=16
{{{1.75, 2.25}, {1.8750000000000016, 2.2499999999999991}, {2.0039062499999982, 2.2460937500000004}, {2.1357421875, 2.2392578125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect11">
{{{1.7535085895385742, 2.4559603499780276}, {1.7508274956738286, 2.3771375952441409}, {1.7496028969921869, 2.2959091083984369}, {1.7501136332812499, 2.2131114089062502}}} id=9
{{{1.3876953125, 2.2373046875}, {1.50390625, 2.24609375}, {1.625, 2.25}, {1.75, 2.25}}} id=16
{{{1.75, 2.25}, {1.8750000000000016, 2.2499999999999991}, {2.0039062499999982, 2.2460937500000004}, {2.1357421875, 2.2392578125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect12">
{{{1.7535085895385742, 2.4559603499780276}, {1.7508274956738286, 2.3771375952441409}, {1.7496028969921869, 2.2959091083984369}, {1.7501136332812499, 2.2131114089062502}}} id=9
{{{1.3876953125, 2.2373046875}, {1.50390625, 2.24609375}, {1.625, 2.25}, {1.75, 2.25}}} id=16
{{{1.75, 2.25}, {1.8124999999999989, 2.2499999999999996}, {1.8759765625000011, 2.2490234374999996}, {1.9403076171875, 2.2471923828125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect13">
{{{1.7535085895385742, 2.4559603499780276}, {1.7508274956738286, 2.3771375952441409}, {1.7496028969921869, 2.2959091083984369}, {1.7501136332812499, 2.2131114089062502}}} id=9
{{{1.5655517578125, 2.2469482421875}, {1.6259765625, 2.2490234375}, {1.6875, 2.25}, {1.75, 2.25}}} id=20
{{{1.75, 2.25}, {1.8124999999999989, 2.2499999999999996}, {1.8759765625000011, 2.2490234374999996}, {1.9403076171875, 2.2471923828125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect14">
{{{1.7506141751022339, 2.3360264837265015}, {1.7500367307348632, 2.2955168052368169}, {1.7498582651367187, 2.2545102586523438}, {1.7501136332812499, 2.2131114089062502}}} id=11
{{{1.5655517578125, 2.2469482421875}, {1.6259765625, 2.2490234375}, {1.6875, 2.25}, {1.75, 2.25}}} id=20
{{{1.75, 2.25}, {1.8124999999999989, 2.2499999999999996}, {1.8759765625000011, 2.2490234374999996}, {1.9403076171875, 2.2471923828125}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect15">
{{{1.7506141751022339, 2.3360264837265015}, {1.7500367307348632, 2.2955168052368169}, {1.7498582651367187, 2.2545102586523438}, {1.7501136332812499, 2.2131114089062502}}} id=11
{{{1.5655517578125, 2.2469482421875}, {1.6259765625, 2.2490234375}, {1.6875, 2.25}, {1.75, 2.25}}} id=20
{{{1.75, 2.25}, {1.7812500000000011, 2.2500000000000004}, {1.8127441406249989, 2.2497558593749991}, {1.8444671630859375, 2.2492828369140625}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect16">
{{{1.7506141751022339, 2.3360264837265015}, {1.7500367307348632, 2.2955168052368169}, {1.7498582651367187, 2.2545102586523438}, {1.7501136332812499, 2.2131114089062502}}} id=11
{{{1.6569976806640625, 2.2492523193359375}, {1.687744140625, 2.249755859375}, {1.71875, 2.25}, {1.75, 2.25}}} id=24
{{{1.75, 2.25}, {1.7812500000000011, 2.2500000000000004}, {1.8127441406249989, 2.2497558593749991}, {1.8444671630859375, 2.2492828369140625}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect17">
{{{1.7500515994997787, 2.274902385537529}, {1.7499667235723877, 2.2544121828619379}, {1.7499859492089846, 2.2338108337792986}, {1.7501136332812499, 2.2131114089062502}}} id=13
{{{1.6569976806640625, 2.2492523193359375}, {1.687744140625, 2.249755859375}, {1.71875, 2.25}, {1.75, 2.25}}} id=24
{{{1.75, 2.25}, {1.7812500000000011, 2.2500000000000004}, {1.8127441406249989, 2.2497558593749991}, {1.8444671630859375, 2.2492828369140625}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect18">
{{{1.7500515994997787, 2.274902385537529}, {1.7499667235723877, 2.2544121828619379}, {1.7499859492089846, 2.2338108337792986}, {1.7501136332812499, 2.2131114089062502}}} id=13
{{{1.6569976806640625, 2.2492523193359375}, {1.687744140625, 2.249755859375}, {1.71875, 2.25}, {1.75, 2.25}}} id=24
{{{1.75, 2.25}, {1.7656249999999989, 2.25}, {1.7813110351562511, 2.24993896484375}, {1.7970561981201172, 2.2498188018798828}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect19">
{{{1.7500515994997787, 2.274902385537529}, {1.7499667235723877, 2.2544121828619379}, {1.7499859492089846, 2.2338108337792986}, {1.7501136332812499, 2.2131114089062502}}} id=13
{{{1.7033100128173828, 2.2498149871826172}, {1.7188110351562504, 2.2499389648437504}, {1.7343749999999991, 2.2499999999999991}, {1.75, 2.25}}} id=28
{{{1.75, 2.25}, {1.7656249999999989, 2.25}, {1.7813110351562511, 2.24993896484375}, {1.7970561981201172, 2.2498188018798828}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect20">
{{{1.7500515994997787, 2.274902385537529}, {1.7500091615360831, 2.2646572841997337}, {1.7499927489633849, 2.2543843962601757}, {1.7500029063906433, 2.2440853555459359}}} id=13
{{{1.7033100128173828, 2.2498149871826172}, {1.7188110351562504, 2.2499389648437504}, {1.7343749999999991, 2.2499999999999991}, {1.75, 2.25}}} id=28
{{{1.75, 2.25}, {1.7656249999999989, 2.25}, {1.7813110351562511, 2.24993896484375}, {1.7970561981201172, 2.2498188018798828}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect21">
{{{1.7500515994997787, 2.274902385537529}, {1.7500091615360831, 2.2646572841997337}, {1.7499927489633849, 2.2543843962601757}, {1.7500029063906433, 2.2440853555459359}}} id=13
{{{1.7033100128173828, 2.2498149871826172}, {1.7188110351562504, 2.2499389648437504}, {1.7343749999999991, 2.2499999999999991}, {1.75, 2.25}}} id=28
{{{1.75, 2.25}, {1.7578125000000004, 2.25}, {1.7656402587890616, 2.2499847412109375}, {1.7734830379486084, 2.2499544620513916}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect22">
{{{1.7500515994997787, 2.274902385537529}, {1.7500091615360831, 2.2646572841997337}, {1.7499927489633849, 2.2543843962601757}, {1.7500029063906433, 2.2440853555459359}}} id=13
{{{1.7266085147857666, 2.2499539852142334}, {1.7343902587890614, 2.2499847412109362}, {1.7421875000000011, 2.2500000000000013}, {1.75, 2.25}}} id=32
{{{1.75, 2.25}, {1.7578125000000004, 2.25}, {1.7656402587890616, 2.2499847412109375}, {1.7734830379486084, 2.2499544620513916}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect23">
{{{1.7500075296736033, 2.2595140978078994}, {1.749999391463374, 2.2543778580665053}, {1.7499978276770138, 2.2492348759030554}, {1.7500029063906433, 2.2440853555459359}}} id=17
{{{1.7266085147857666, 2.2499539852142334}, {1.7343902587890614, 2.2499847412109362}, {1.7421875000000011, 2.2500000000000013}, {1.75, 2.25}}} id=32
{{{1.75, 2.25}, {1.7578125000000004, 2.25}, {1.7656402587890616, 2.2499847412109375}, {1.7734830379486084, 2.2499544620513916}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect24">
{{{1.7500075296736033, 2.2595140978078994}, {1.749999391463374, 2.2543778580665053}, {1.7499978276770138, 2.2492348759030554}, {1.7500029063906433, 2.2440853555459359}}} id=17
{{{1.7266085147857666, 2.2499539852142334}, {1.7343902587890614, 2.2499847412109362}, {1.7421875000000011, 2.2500000000000013}, {1.75, 2.25}}} id=32
{{{1.75, 2.25}, {1.7539062499999991, 2.25}, {1.7578163146972661, 2.2499961853027344}, {1.7617301642894745, 2.2499885857105255}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect25">
{{{1.7500075296736033, 2.2595140978078994}, {1.749999391463374, 2.2543778580665053}, {1.7499978276770138, 2.2492348759030554}, {1.7500029063906433, 2.2440853555459359}}} id=17
{{{1.7382927238941193, 2.2499885261058807}, {1.7421913146972665, 2.2499961853027353}, {1.7460937499999996, 2.2499999999999996}, {1.75, 2.25}}} id=36
{{{1.75, 2.25}, {1.7539062499999991, 2.25}, {1.7578163146972661, 2.2499961853027344}, {1.7617301642894745, 2.2499885857105255}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect26">
{{{1.7500002616856762, 2.2518047069078144}, {1.7499994883020116, 2.2492332413546396}, {1.7500003670338282, 2.2466601157244943}, {1.7500029063906433, 2.2440853555459359}}} id=19
{{{1.75, 2.25}, {1.7539062499999991, 2.25}, {1.7578163146972661, 2.2499961853027344}, {1.7617301642894745, 2.2499885857105255}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect27">
{{{1.7500002616856762, 2.2518047069078144}, {1.7499994883020116, 2.2492332413546396}, {1.7500003670338282, 2.2466601157244943}, {1.7500029063906433, 2.2440853555459359}}} id=19
{{{1.75, 2.25}, {1.7519531250000011, 2.2499999999999991}, {1.7539072036743153, 2.249999046325684}, {1.7558622322976589, 2.2499971427023411}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect28">
{{{1.7500002616856762, 2.2518047069078144}, {1.749999874993843, 2.2505189741312273}, {1.7499999013308822, 2.2492328263353962}, {1.7500003417604797, 2.2479462667113941}}} id=19
{{{1.75, 2.25}, {1.7519531250000011, 2.2499999999999991}, {1.7539072036743153, 2.249999046325684}, {1.7558622322976589, 2.2499971427023411}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect29">
{{{1.7500002616856762, 2.2518047069078144}, {1.749999874993843, 2.2505189741312273}, {1.7499999013308822, 2.2492328263353962}, {1.7500003417604797, 2.2479462667113941}}} id=19
{{{1.75, 2.25}, {1.7509765624999989, 2.2499999999999996}, {1.7519533634185802, 2.2499997615814205}, {1.752930402290076, 2.249999285209924}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect30">
{{{1.7500002616856762, 2.2518047069078144}, {1.7500000683397601, 2.2511618405195208}, {1.7499999782510609, 2.2505188703764163}, {1.7499999915525417, 2.2498757968773848}}} id=19
{{{1.75, 2.25}, {1.7509765624999989, 2.2499999999999996}, {1.7519533634185802, 2.2499997615814205}, {1.752930402290076, 2.249999285209924}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect31">
{{{1.7500002616856762, 2.2518047069078144}, {1.7500000683397601, 2.2511618405195208}, {1.7499999782510609, 2.2505188703764163}, {1.7499999915525417, 2.2498757968773848}}} id=19
{{{1.75, 2.25}, {1.7504882812500011, 2.2500000000000004}, {1.7509766221046437, 2.2499999403953543}, {1.7514650225057267, 2.2499998212442733}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect32">
{{{1.7500000491263352, 2.2508403295591259}, {1.7500000040986061, 2.2505188445374351}, {1.7499999849018013, 2.2501973336269003}, {1.7499999915525417, 2.2498757968773848}}} id=25
{{{1.75, 2.25}, {1.7504882812500011, 2.2500000000000004}, {1.7509766221046437, 2.2499999403953543}, {1.7514650225057267, 2.2499998212442733}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect33">
{{{1.7500000491263352, 2.2508403295591259}, {1.7500000040986061, 2.2505188445374351}, {1.7499999849018013, 2.2501973336269003}, {1.7499999915525417, 2.2498757968773848}}} id=25
{{{1.75, 2.25}, {1.7502441406249989, 2.25}, {1.7504882961511623, 2.2499999850988388}, {1.7507324665712076, 2.2499999553037924}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect34">
{{{1.7500000009600125, 2.2503580826161893}, {1.7499999913636877, 2.2501973271671556}, {1.7499999882271713, 2.2500365652521426}, {1.7499999915525417, 2.2498757968773848}}} id=27
{{{1.75, 2.25}, {1.7502441406249989, 2.25}, {1.7504882961511623, 2.2499999850988388}, {1.7507324665712076, 2.2499999553037924}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect35">
{{{1.7500000009600125, 2.2503580826161893}, {1.7499999913636877, 2.2501973271671556}, {1.7499999882271713, 2.2500365652521426}, {1.7499999915525417, 2.2498757968773848}}} id=27
{{{1.75, 2.25}, {1.7501220703125004, 2.25}, {1.7502441443502894, 2.2499999962747097}, {1.7503662221124614, 2.2499999888250386}}} id=4
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<div id="sect36">
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
</div>
</div>
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
<script type="text/javascript">
These tests stress pathops by describing the union of circle-like paths that have tiny line segments embedded and double back to create near-coincident conditions. The fixes include - detect when finding the active top loops between two possible answers - preflight chasing winding to ensure answer is consistent - binary search more often when quadratic intersection fails - add more failure paths when an intersect is missed While this fixes the chrome bug, reenabling path ops in svg should be deferred until additional fixes are landed. TBR= BUG=421132 Committed: https://skia.googlesource.com/skia/+/6f726addf3178b01949bb389ef83cf14a1d7b6b2 Review URL: https://codereview.chromium.org/633393002
2014-11-13 14:58:52 +00:00
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var testDivs = [
sect1,
sect2,
sect3,
sect4,
sect5,
sect6,
sect7,
sect8,
sect9,
sect10,
sect11,
sect12,
sect13,
sect14,
sect15,
sect16,
sect17,
sect18,
sect19,
sect20,
sect21,
sect22,
sect23,
sect24,
sect25,
sect26,
sect27,
sect28,
sect29,
sect30,
sect31,
sect32,
sect33,
sect34,
sect35,
sect36,
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
];
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var decimal_places = 3;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var tests = [];
var testTitles = [];
var testIndex = 0;
var ctx;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var subscale = 1;
var xmin, xmax, ymin, ymax;
var scale;
var initScale;
var mouseX, mouseY;
var mouseDown = false;
var srcLeft, srcTop;
var screenWidth, screenHeight;
var drawnPts;
var curveT = 0;
var lastX, lastY;
var activeCurve = [];
var activePt;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var ids = [];
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var focus_on_selection = 0;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var draw_t = false;
var draw_closest_t = false;
add cubic red option to pathops tool R=reed@google.com TBR=reed@google.com NOTREECHECKS=true NOTRY=true Author: caryclark@google.com Review URL: https://codereview.chromium.org/428963004
2014-07-29 19:13:28 +00:00
var draw_cubic_red = false;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var draw_derivative = false;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var draw_endpoints = 2;
var draw_id = false;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var draw_midpoint = 0;
var draw_mouse_xy = false;
var draw_order = false;
var draw_point_xy = false;
var draw_ray_intersect = false;
var draw_quarterpoint = 0;
var draw_tangents = 1;
var draw_sortpoint = 0;
var retina_scale = !!window.devicePixelRatio;
function parse(test, title) {
var curveStrs = test.split("{{");
var pattern = /-?\d+\.*\d*e?-?\d*/g;
var curves = [];
for (var c in curveStrs) {
var curveStr = curveStrs[c];
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var idPart = curveStr.split("id=");
var id = -1;
if (idPart.length == 2) {
id = parseInt(idPart[1]);
curveStr = idPart[0];
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var points = curveStr.match(pattern);
var pts = [];
for (var wd in points) {
var num = parseFloat(points[wd]);
if (isNaN(num)) continue;
pts.push(num);
}
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (pts.length > 2) {
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
curves.push(pts);
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
}
if (id >= 0) {
ids.push(id);
ids.push(pts);
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
}
if (curves.length >= 1) {
tests.push(curves);
testTitles.push(title);
}
}
function init(test) {
var canvas = document.getElementById('canvas');
if (!canvas.getContext) return;
ctx = canvas.getContext('2d');
var resScale = retina_scale && window.devicePixelRatio ? window.devicePixelRatio : 1;
var unscaledWidth = window.innerWidth - 20;
var unscaledHeight = window.innerHeight - 20;
screenWidth = unscaledWidth;
screenHeight = unscaledHeight;
canvas.width = unscaledWidth * resScale;
canvas.height = unscaledHeight * resScale;
canvas.style.width = unscaledWidth + 'px';
canvas.style.height = unscaledHeight + 'px';
if (resScale != 1) {
ctx.scale(resScale, resScale);
}
xmin = Infinity;
xmax = -Infinity;
ymin = Infinity;
ymax = -Infinity;
for (var curves in test) {
var curve = test[curves];
var last = curve.length;
for (var idx = 0; idx < last; idx += 2) {
xmin = Math.min(xmin, curve[idx]);
xmax = Math.max(xmax, curve[idx]);
ymin = Math.min(ymin, curve[idx + 1]);
ymax = Math.max(ymax, curve[idx + 1]);
}
}
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
xmin -= Math.min(1, Math.max(xmax - xmin, ymax - ymin));
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var testW = xmax - xmin;
var testH = ymax - ymin;
subscale = 1;
while (testW * subscale < 0.1 && testH * subscale < 0.1) {
subscale *= 10;
}
while (testW * subscale > 10 && testH * subscale > 10) {
subscale /= 10;
}
setScale(xmin, xmax, ymin, ymax);
mouseX = (screenWidth / 2) / scale + srcLeft;
mouseY = (screenHeight / 2) / scale + srcTop;
initScale = scale;
}
function setScale(x0, x1, y0, y1) {
var srcWidth = x1 - x0;
var srcHeight = y1 - y0;
var usableWidth = screenWidth;
var xDigits = Math.ceil(Math.log(Math.abs(xmax)) / Math.log(10));
var yDigits = Math.ceil(Math.log(Math.abs(ymax)) / Math.log(10));
usableWidth -= (xDigits + yDigits) * 10;
usableWidth -= decimal_places * 10;
var hscale = usableWidth / srcWidth;
var vscale = screenHeight / srcHeight;
scale = Math.min(hscale, vscale);
var invScale = 1 / scale;
var sxmin = x0 - invScale * 5;
var symin = y0 - invScale * 10;
var sxmax = x1 + invScale * (6 * decimal_places + 10);
var symax = y1 + invScale * 10;
srcWidth = sxmax - sxmin;
srcHeight = symax - symin;
hscale = usableWidth / srcWidth;
vscale = screenHeight / srcHeight;
scale = Math.min(hscale, vscale);
srcLeft = sxmin;
srcTop = symin;
}
function dxy_at_t(curve, t) {
var dxy = {};
if (curve.length == 6) {
var a = t - 1;
var b = 1 - 2 * t;
var c = t;
dxy.x = a * curve[0] + b * curve[2] + c * curve[4];
dxy.y = a * curve[1] + b * curve[3] + c * curve[5];
} else if (curve.length == 8) {
var one_t = 1 - t;
var a = curve[0];
var b = curve[2];
var c = curve[4];
var d = curve[6];
dxy.x = 3 * ((b - a) * one_t * one_t + 2 * (c - b) * t * one_t + (d - c) * t * t);
a = curve[1];
b = curve[3];
c = curve[5];
d = curve[7];
dxy.y = 3 * ((b - a) * one_t * one_t + 2 * (c - b) * t * one_t + (d - c) * t * t);
}
return dxy;
}
var flt_epsilon = 1.19209290E-07;
function approximately_zero(A) {
return Math.abs(A) < flt_epsilon;
}
function approximately_zero_inverse(A) {
return Math.abs(A) > (1 / flt_epsilon);
}
function quad_real_roots(A, B, C) {
var s = [];
var p = B / (2 * A);
var q = C / A;
if (approximately_zero(A) && (approximately_zero_inverse(p)
|| approximately_zero_inverse(q))) {
if (approximately_zero(B)) {
if (C == 0) {
s[0] = 0;
}
return s;
}
s[0] = -C / B;
return s;
}
/* normal form: x^2 + px + q = 0 */
var p2 = p * p;
if (!approximately_zero(p2 - q) && p2 < q) {
return s;
}
var sqrt_D = 0;
if (p2 > q) {
sqrt_D = Math.sqrt(p2 - q);
}
s[0] = sqrt_D - p;
var flip = -sqrt_D - p;
if (!approximately_zero(s[0] - flip)) {
s[1] = flip;
}
return s;
}
function cubic_real_roots(A, B, C, D) {
if (approximately_zero(A)) { // we're just a quadratic
return quad_real_roots(B, C, D);
}
if (approximately_zero(D)) { // 0 is one root
var s = quad_real_roots(A, B, C);
for (var i = 0; i < s.length; ++i) {
if (approximately_zero(s[i])) {
return s;
}
}
s.push(0);
return s;
}
if (approximately_zero(A + B + C + D)) { // 1 is one root
var s = quad_real_roots(A, A + B, -D);
for (var i = 0; i < s.length; ++i) {
if (approximately_zero(s[i] - 1)) {
return s;
}
}
s.push(1);
return s;
}
var a, b, c;
var invA = 1 / A;
a = B * invA;
b = C * invA;
c = D * invA;
var a2 = a * a;
var Q = (a2 - b * 3) / 9;
var R = (2 * a2 * a - 9 * a * b + 27 * c) / 54;
var R2 = R * R;
var Q3 = Q * Q * Q;
var R2MinusQ3 = R2 - Q3;
var adiv3 = a / 3;
var r;
var roots = [];
if (R2MinusQ3 < 0) { // we have 3 real roots
var theta = Math.acos(R / Math.sqrt(Q3));
var neg2RootQ = -2 * Math.sqrt(Q);
r = neg2RootQ * Math.cos(theta / 3) - adiv3;
roots.push(r);
r = neg2RootQ * Math.cos((theta + 2 * Math.PI) / 3) - adiv3;
if (!approximately_zero(roots[0] - r)) {
roots.push(r);
}
r = neg2RootQ * Math.cos((theta - 2 * Math.PI) / 3) - adiv3;
if (!approximately_zero(roots[0] - r) && (roots.length == 1
|| !approximately_zero(roots[1] - r))) {
roots.push(r);
}
} else { // we have 1 real root
var sqrtR2MinusQ3 = Math.sqrt(R2MinusQ3);
var A = Math.abs(R) + sqrtR2MinusQ3;
A = Math.pow(A, 1/3);
if (R > 0) {
A = -A;
}
if (A != 0) {
A += Q / A;
}
r = A - adiv3;
roots.push(r);
if (approximately_zero(R2 - Q3)) {
r = -A / 2 - adiv3;
if (!approximately_zero(s[0] - r)) {
roots.push(r);
}
}
}
return roots;
}
function approximately_zero_or_more(tValue) {
return tValue >= -flt_epsilon;
}
function approximately_one_or_less(tValue) {
return tValue <= 1 + flt_epsilon;
}
function approximately_less_than_zero(tValue) {
return tValue < flt_epsilon;
}
function approximately_greater_than_one(tValue) {
return tValue > 1 - flt_epsilon;
}
function add_valid_ts(s) {
var t = [];
nextRoot:
for (var index = 0; index < s.length; ++index) {
var tValue = s[index];
if (approximately_zero_or_more(tValue) && approximately_one_or_less(tValue)) {
if (approximately_less_than_zero(tValue)) {
tValue = 0;
} else if (approximately_greater_than_one(tValue)) {
tValue = 1;
}
for (var idx2 = 0; idx2 < t.length; ++idx2) {
if (approximately_zero(t[idx2] - tValue)) {
continue nextRoot;
}
}
t.push(tValue);
}
}
return t;
}
function quad_roots(A, B, C) {
var s = quad_real_roots(A, B, C);
var foundRoots = add_valid_ts(s);
return foundRoots;
}
function cubic_roots(A, B, C, D) {
var s = cubic_real_roots(A, B, C, D);
var foundRoots = add_valid_ts(s);
return foundRoots;
}
function ray_curve_intersect(startPt, endPt, curve) {
var adj = endPt[0] - startPt[0];
var opp = endPt[1] - startPt[1];
var r = [];
for (var n = 0; n < curve.length / 2; ++n) {
r[n] = (curve[n * 2 + 1] - startPt[1]) * adj - (curve[n * 2] - startPt[0]) * opp;
}
if (curve.length == 6) {
var A = r[2];
var B = r[1];
var C = r[0];
A += C - 2 * B; // A = a - 2*b + c
B -= C; // B = -(b - c)
return quad_roots(A, 2 * B, C);
}
var A = r[3]; // d
var B = r[2] * 3; // 3*c
var C = r[1] * 3; // 3*b
var D = r[0]; // a
A -= D - C + B; // A = -a + 3*b - 3*c + d
B += 3 * D - 2 * C; // B = 3*a - 6*b + 3*c
C -= 3 * D; // C = -3*a + 3*b
return cubic_roots(A, B, C, D);
}
function x_at_t(curve, t) {
var one_t = 1 - t;
if (curve.length == 4) {
return one_t * curve[0] + t * curve[2];
}
var one_t2 = one_t * one_t;
var t2 = t * t;
if (curve.length == 6) {
return one_t2 * curve[0] + 2 * one_t * t * curve[2] + t2 * curve[4];
}
var a = one_t2 * one_t;
var b = 3 * one_t2 * t;
var c = 3 * one_t * t2;
var d = t2 * t;
return a * curve[0] + b * curve[2] + c * curve[4] + d * curve[6];
}
function y_at_t(curve, t) {
var one_t = 1 - t;
if (curve.length == 4) {
return one_t * curve[1] + t * curve[3];
}
var one_t2 = one_t * one_t;
var t2 = t * t;
if (curve.length == 6) {
return one_t2 * curve[1] + 2 * one_t * t * curve[3] + t2 * curve[5];
}
var a = one_t2 * one_t;
var b = 3 * one_t2 * t;
var c = 3 * one_t * t2;
var d = t2 * t;
return a * curve[1] + b * curve[3] + c * curve[5] + d * curve[7];
}
function drawPointAtT(curve) {
var x = x_at_t(curve, curveT);
var y = y_at_t(curve, curveT);
drawPoint(x, y);
}
function drawLine(x1, y1, x2, y2) {
ctx.beginPath();
ctx.moveTo((x1 - srcLeft) * scale,
(y1 - srcTop) * scale);
ctx.lineTo((x2 - srcLeft) * scale,
(y2 - srcTop) * scale);
ctx.stroke();
}
function drawPoint(px, py) {
for (var pts = 0; pts < drawnPts.length; pts += 2) {
var x = drawnPts[pts];
var y = drawnPts[pts + 1];
if (px == x && py == y) {
return;
}
}
drawnPts.push(px);
drawnPts.push(py);
var _px = (px - srcLeft) * scale;
var _py = (py - srcTop) * scale;
ctx.beginPath();
ctx.arc(_px, _py, 3, 0, Math.PI * 2, true);
ctx.closePath();
ctx.stroke();
if (draw_point_xy) {
var label = px.toFixed(decimal_places) + ", " + py.toFixed(decimal_places);
ctx.font = "normal 10px Arial";
ctx.textAlign = "left";
ctx.fillStyle = "black";
ctx.fillText(label, _px + 5, _py);
}
}
function drawPointSolid(px, py) {
drawPoint(px, py);
ctx.fillStyle = "rgba(0,0,0, 0.4)";
ctx.fill();
}
function crossPt(origin, pt1, pt2) {
return ((pt1[0] - origin[0]) * (pt2[1] - origin[1])
- (pt1[1] - origin[1]) * (pt2[0] - origin[0])) > 0 ? 0 : 1;
}
// may not work well for cubics
function curveClosestT(curve, x, y) {
var closest = -1;
var closestDist = Infinity;
var l = Infinity, t = Infinity, r = -Infinity, b = -Infinity;
for (var i = 0; i < 16; ++i) {
var testX = x_at_t(curve, i / 16);
l = Math.min(testX, l);
r = Math.max(testX, r);
var testY = y_at_t(curve, i / 16);
t = Math.min(testY, t);
b = Math.max(testY, b);
var dx = testX - x;
var dy = testY - y;
var dist = dx * dx + dy * dy;
if (closestDist > dist) {
closestDist = dist;
closest = i;
}
}
var boundsX = r - l;
var boundsY = b - t;
var boundsDist = boundsX * boundsX + boundsY * boundsY;
if (closestDist > boundsDist) {
return -1;
}
console.log("closestDist = " + closestDist + " boundsDist = " + boundsDist
+ " t = " + closest / 16);
return closest / 16;
}
function draw(test, title) {
ctx.font = "normal 50px Arial";
ctx.textAlign = "left";
ctx.fillStyle = "rgba(0,0,0, 0.1)";
ctx.fillText(title, 50, 50);
ctx.font = "normal 10px Arial";
// ctx.lineWidth = "1.001"; "0.999";
var hullStarts = [];
var hullEnds = [];
var midSpokes = [];
var midDist = [];
var origin = [];
var shortSpokes = [];
var shortDist = [];
var sweeps = [];
drawnPts = [];
for (var curves in test) {
var curve = test[curves];
origin.push(curve[0]);
origin.push(curve[1]);
var startPt = [];
startPt.push(curve[2]);
startPt.push(curve[3]);
hullStarts.push(startPt);
var endPt = [];
if (curve.length == 4) {
endPt.push(curve[2]);
endPt.push(curve[3]);
} else if (curve.length == 6) {
endPt.push(curve[4]);
endPt.push(curve[5]);
} else if (curve.length == 8) {
endPt.push(curve[6]);
endPt.push(curve[7]);
}
hullEnds.push(endPt);
var sweep = crossPt(origin, startPt, endPt);
sweeps.push(sweep);
var midPt = [];
midPt.push(x_at_t(curve, 0.5));
midPt.push(y_at_t(curve, 0.5));
midSpokes.push(midPt);
var shortPt = [];
shortPt.push(x_at_t(curve, 0.25));
shortPt.push(y_at_t(curve, 0.25));
shortSpokes.push(shortPt);
var dx = midPt[0] - origin[0];
var dy = midPt[1] - origin[1];
var dist = Math.sqrt(dx * dx + dy * dy);
midDist.push(dist);
dx = shortPt[0] - origin[0];
dy = shortPt[1] - origin[1];
dist = Math.sqrt(dx * dx + dy * dy);
shortDist.push(dist);
}
var intersect = [];
var useIntersect = false;
var maxWidth = Math.max(xmax - xmin, ymax - ymin);
for (var curves in test) {
var curve = test[curves];
if (curve.length == 6 || curve.length == 8) {
fix minor skp-found bugs remove globals from pathops_unittest BUG=skia:2460 TBR=mtklein Author: caryclark@google.com Review URL: https://codereview.chromium.org/239563004 git-svn-id: http://skia.googlecode.com/svn/trunk@14378 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-25 12:59:11 +00:00
var opp = curves == 0 || curves == 1 ? 0 : 1;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var sects = ray_curve_intersect(origin, hullEnds[opp], curve);
intersect.push(sects);
if (sects.length > 1) {
var intersection = sects[0];
if (intersection == 0) {
intersection = sects[1];
}
var ix = x_at_t(curve, intersection) - origin[0];
var iy = y_at_t(curve, intersection) - origin[1];
var ex = hullEnds[opp][0] - origin[0];
var ey = hullEnds[opp][1] - origin[1];
if (ix * ex >= 0 && iy * ey >= 0) {
var iDist = Math.sqrt(ix * ix + iy * iy);
var eDist = Math.sqrt(ex * ex + ey * ey);
var delta = Math.abs(iDist - eDist) / maxWidth;
if (delta > (curve.length == 6 ? 1e-5 : 1e-4)) {
useIntersect ^= true;
}
}
}
}
}
fix minor skp-found bugs remove globals from pathops_unittest BUG=skia:2460 TBR=mtklein Author: caryclark@google.com Review URL: https://codereview.chromium.org/239563004 git-svn-id: http://skia.googlecode.com/svn/trunk@14378 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-25 12:59:11 +00:00
var midLeft = curves != 0 ? crossPt(origin, midSpokes[0], midSpokes[1]) : 0;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
var firstInside;
if (useIntersect) {
var sect1 = intersect[0].length > 1;
var sIndex = sect1 ? 0 : 1;
var sects = intersect[sIndex];
var intersection = sects[0];
if (intersection == 0) {
intersection = sects[1];
}
var curve = test[sIndex];
var ix = x_at_t(curve, intersection) - origin[0];
var iy = y_at_t(curve, intersection) - origin[1];
var opp = sect1 ? 1 : 0;
var ex = hullEnds[opp][0] - origin[0];
var ey = hullEnds[opp][1] - origin[1];
var iDist = ix * ix + iy * iy;
var eDist = ex * ex + ey * ey;
firstInside = (iDist > eDist) ^ (sIndex == 0) ^ sweeps[0];
// console.log("iDist=" + iDist + " eDist=" + eDist + " sIndex=" + sIndex
// + " sweeps[0]=" + sweeps[0]);
} else {
// console.log("midLeft=" + midLeft);
firstInside = midLeft != 0;
}
var shorter = midDist[1] < midDist[0];
var shortLeft = shorter ? crossPt(origin, shortSpokes[0], midSpokes[1])
: crossPt(origin, midSpokes[0], shortSpokes[1]);
var startCross = crossPt(origin, hullStarts[0], hullStarts[1]);
var disallowShort = midLeft == startCross && midLeft == sweeps[0]
&& midLeft == sweeps[1];
// console.log("midLeft=" + midLeft + " startCross=" + startCross);
var intersectIndex = 0;
for (var curves in test) {
var curve = test[curves];
if (curve.length != 4 && curve.length != 6 && curve.length != 8) {
continue;
}
ctx.lineWidth = 1;
if (draw_tangents != 0) {
add cubic red option to pathops tool R=reed@google.com TBR=reed@google.com NOTREECHECKS=true NOTRY=true Author: caryclark@google.com Review URL: https://codereview.chromium.org/428963004
2014-07-29 19:13:28 +00:00
if (draw_cubic_red ? curve.length == 8 : firstInside == curves) {
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
ctx.strokeStyle = "rgba(255,0,0, 0.3)";
} else {
ctx.strokeStyle = "rgba(0,0,255, 0.3)";
}
drawLine(curve[0], curve[1], curve[2], curve[3]);
if (draw_tangents != 2) {
if (curve.length > 4) drawLine(curve[2], curve[3], curve[4], curve[5]);
if (curve.length > 6) drawLine(curve[4], curve[5], curve[6], curve[7]);
}
if (draw_tangents != 1) {
if (curve.length == 6) drawLine(curve[0], curve[1], curve[4], curve[5]);
if (curve.length == 8) drawLine(curve[0], curve[1], curve[6], curve[7]);
}
}
ctx.beginPath();
ctx.moveTo((curve[0] - srcLeft) * scale, (curve[1] - srcTop) * scale);
if (curve.length == 4) {
ctx.lineTo((curve[2] - srcLeft) * scale, (curve[3] - srcTop) * scale);
} else if (curve.length == 6) {
ctx.quadraticCurveTo(
(curve[2] - srcLeft) * scale, (curve[3] - srcTop) * scale,
(curve[4] - srcLeft) * scale, (curve[5] - srcTop) * scale);
} else {
ctx.bezierCurveTo(
(curve[2] - srcLeft) * scale, (curve[3] - srcTop) * scale,
(curve[4] - srcLeft) * scale, (curve[5] - srcTop) * scale,
(curve[6] - srcLeft) * scale, (curve[7] - srcTop) * scale);
}
add cubic red option to pathops tool R=reed@google.com TBR=reed@google.com NOTREECHECKS=true NOTRY=true Author: caryclark@google.com Review URL: https://codereview.chromium.org/428963004
2014-07-29 19:13:28 +00:00
if (draw_cubic_red ? curve.length == 8 : firstInside == curves) {
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
ctx.strokeStyle = "rgba(255,0,0, 1)";
} else {
ctx.strokeStyle = "rgba(0,0,255, 1)";
}
ctx.stroke();
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (draw_endpoints > 0) {
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
drawPoint(curve[0], curve[1]);
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (draw_endpoints > 1 || curve.length == 4) {
drawPoint(curve[2], curve[3]);
}
if (curve.length == 6 || (draw_endpoints > 1 && curve.length == 8)) {
drawPoint(curve[4], curve[5]);
}
if (curve.length == 8) drawPoint(curve[6], curve[7]);
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
}
if (draw_midpoint != 0) {
if ((curves == 0) == (midLeft == 0)) {
ctx.strokeStyle = "rgba(0,180,127, 0.6)";
} else {
ctx.strokeStyle = "rgba(127,0,127, 0.6)";
}
var midX = x_at_t(curve, 0.5);
var midY = y_at_t(curve, 0.5);
drawPointSolid(midX, midY);
if (draw_midpoint > 1) {
drawLine(curve[0], curve[1], midX, midY);
}
}
if (draw_quarterpoint != 0) {
if ((curves == 0) == (shortLeft == 0)) {
ctx.strokeStyle = "rgba(0,191,63, 0.6)";
} else {
ctx.strokeStyle = "rgba(63,0,191, 0.6)";
}
var midT = (curves == 0) == shorter ? 0.25 : 0.5;
var midX = x_at_t(curve, midT);
var midY = y_at_t(curve, midT);
drawPointSolid(midX, midY);
if (draw_quarterpoint > 1) {
drawLine(curve[0], curve[1], midX, midY);
}
}
if (draw_sortpoint != 0) {
if ((curves == 0) == ((disallowShort == -1 ? midLeft : shortLeft) == 0)) {
ctx.strokeStyle = "rgba(0,155,37, 0.6)";
} else {
ctx.strokeStyle = "rgba(37,0,155, 0.6)";
}
var midT = (curves == 0) == shorter && disallowShort != curves ? 0.25 : 0.5;
console.log("curves=" + curves + " disallowShort=" + disallowShort
+ " midLeft=" + midLeft + " shortLeft=" + shortLeft
+ " shorter=" + shorter + " midT=" + midT);
var midX = x_at_t(curve, midT);
var midY = y_at_t(curve, midT);
drawPointSolid(midX, midY);
if (draw_sortpoint > 1) {
drawLine(curve[0], curve[1], midX, midY);
}
}
if (draw_ray_intersect != 0) {
ctx.strokeStyle = "rgba(75,45,199, 0.6)";
if (curve.length == 6 || curve.length == 8) {
var intersections = intersect[intersectIndex];
for (var i in intersections) {
var intersection = intersections[i];
var x = x_at_t(curve, intersection);
var y = y_at_t(curve, intersection);
drawPointSolid(x, y);
if (draw_ray_intersect > 1) {
drawLine(curve[0], curve[1], x, y);
}
}
}
++intersectIndex;
}
if (draw_order) {
var px = x_at_t(curve, 0.75);
var py = y_at_t(curve, 0.75);
var _px = (px - srcLeft) * scale;
var _py = (py - srcTop) * scale;
ctx.beginPath();
ctx.arc(_px, _py, 15, 0, Math.PI * 2, true);
ctx.closePath();
ctx.fillStyle = "white";
ctx.fill();
add cubic red option to pathops tool R=reed@google.com TBR=reed@google.com NOTREECHECKS=true NOTRY=true Author: caryclark@google.com Review URL: https://codereview.chromium.org/428963004
2014-07-29 19:13:28 +00:00
if (draw_cubic_red ? curve.length == 8 : firstInside == curves) {
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
ctx.strokeStyle = "rgba(255,0,0, 1)";
ctx.fillStyle = "rgba(255,0,0, 1)";
} else {
ctx.strokeStyle = "rgba(0,0,255, 1)";
ctx.fillStyle = "rgba(0,0,255, 1)";
}
ctx.stroke();
ctx.font = "normal 16px Arial";
ctx.textAlign = "center";
ctx.fillText(parseInt(curves) + 1, _px, _py + 5);
}
if (draw_closest_t) {
var t = curveClosestT(curve, mouseX, mouseY);
if (t >= 0) {
var x = x_at_t(curve, t);
var y = y_at_t(curve, t);
drawPointSolid(x, y);
}
}
if (!approximately_zero(scale - initScale)) {
ctx.font = "normal 20px Arial";
ctx.fillStyle = "rgba(0,0,0, 0.3)";
ctx.textAlign = "right";
ctx.fillText(scale.toFixed(decimal_places) + 'x',
screenWidth - 10, screenHeight - 5);
}
if (draw_t) {
drawPointAtT(curve);
}
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (draw_id) {
var id = -1;
for (var i = 0; i < ids.length; i += 2) {
if (ids[i + 1] == curve) {
id = ids[i];
break;
}
}
if (id >= 0) {
var px = x_at_t(curve, 0.5);
var py = y_at_t(curve, 0.5);
var _px = (px - srcLeft) * scale;
var _py = (py - srcTop) * scale;
ctx.beginPath();
ctx.arc(_px, _py, 15, 0, Math.PI * 2, true);
ctx.closePath();
ctx.fillStyle = "white";
ctx.fill();
ctx.strokeStyle = "rgba(255,0,0, 1)";
ctx.fillStyle = "rgba(255,0,0, 1)";
ctx.stroke();
ctx.font = "normal 16px Arial";
ctx.textAlign = "center";
ctx.fillText(id, _px, _py + 5);
}
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
}
if (draw_t) {
drawCurveTControl();
}
}
function drawCurveTControl() {
ctx.lineWidth = 2;
ctx.strokeStyle = "rgba(0,0,0, 0.3)";
ctx.beginPath();
ctx.rect(screenWidth - 80, 40, 28, screenHeight - 80);
ctx.stroke();
var ty = 40 + curveT * (screenHeight - 80);
ctx.beginPath();
ctx.moveTo(screenWidth - 80, ty);
ctx.lineTo(screenWidth - 85, ty - 5);
ctx.lineTo(screenWidth - 85, ty + 5);
ctx.lineTo(screenWidth - 80, ty);
ctx.fillStyle = "rgba(0,0,0, 0.6)";
ctx.fill();
var num = curveT.toFixed(decimal_places);
ctx.font = "normal 10px Arial";
ctx.textAlign = "left";
ctx.fillText(num, screenWidth - 78, ty);
}
function ptInTControl() {
var e = window.event;
var tgt = e.target || e.srcElement;
var left = tgt.offsetLeft;
var top = tgt.offsetTop;
var x = (e.clientX - left);
var y = (e.clientY - top);
if (x < screenWidth - 80 || x > screenWidth - 50) {
return false;
}
if (y < 40 || y > screenHeight - 80) {
return false;
}
curveT = (y - 40) / (screenHeight - 120);
if (curveT < 0 || curveT > 1) {
throw "stop execution";
}
return true;
}
function drawTop() {
init(tests[testIndex]);
redraw();
}
function redraw() {
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (focus_on_selection > 0) {
var focusXmin = focusYmin = Infinity;
var focusXmax = focusYmax = -Infinity;
var choice = 0;
for (var curves in tests[testIndex]) {
if (++choice != focus_on_selection) {
continue;
}
var curve = tests[testIndex][curves];
var last = curve.length;
for (var idx = 0; idx < last; idx += 2) {
focusXmin = Math.min(focusXmin, curve[idx]);
focusXmax = Math.max(focusXmax, curve[idx]);
focusYmin = Math.min(focusYmin, curve[idx + 1]);
focusYmax = Math.max(focusYmax, curve[idx + 1]);
}
}
focusXmin -= Math.min(1, Math.max(focusXmax - focusXmin, focusYmax - focusYmin));
if (focusXmin < focusXmax && focusYmin < focusYmax) {
setScale(focusXmin, focusXmax, focusYmin, focusYmax);
}
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
ctx.beginPath();
ctx.rect(0, 0, ctx.canvas.width, ctx.canvas.height);
ctx.fillStyle = "white";
ctx.fill();
draw(tests[testIndex], testTitles[testIndex]);
}
function doKeyPress(evt) {
var char = String.fromCharCode(evt.charCode);
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
var focusWasOn = false;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
switch (char) {
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
decimal_places = char - '0';
redraw();
break;
case '-':
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
focusWasOn = focus_on_selection;
if (focusWasOn) {
focus_on_selection = false;
scale /= 1.2;
} else {
scale /= 2;
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
calcLeftTop();
redraw();
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
focus_on_selection = focusWasOn;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
break;
case '=':
case '+':
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
focusWasOn = focus_on_selection;
if (focusWasOn) {
focus_on_selection = false;
scale *= 1.2;
} else {
scale *= 2;
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
calcLeftTop();
redraw();
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
focus_on_selection = focusWasOn;
add cubic red option to pathops tool R=reed@google.com TBR=reed@google.com NOTREECHECKS=true NOTRY=true Author: caryclark@google.com Review URL: https://codereview.chromium.org/428963004
2014-07-29 19:13:28 +00:00
break;
case 'b':
draw_cubic_red ^= true;
redraw();
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
break;
case 'c':
drawTop();
break;
case 'd':
var test = tests[testIndex];
var testClone = [];
for (var curves in test) {
var c = test[curves];
var cClone = [];
for (var index = 0; index < c.length; ++index) {
cClone.push(c[index]);
}
testClone.push(cClone);
}
tests.push(testClone);
testTitles.push(testTitles[testIndex] + " copy");
testIndex = tests.length - 1;
redraw();
break;
case 'e':
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
draw_endpoints = (draw_endpoints + 1) % 3;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
redraw();
break;
case 'f':
draw_derivative ^= true;
redraw();
break;
case 'i':
draw_ray_intersect = (draw_ray_intersect + 1) % 3;
redraw();
break;
case 'l':
var test = tests[testIndex];
console.log("<div id=\"" + testTitles[testIndex] + "\" >");
for (var curves in test) {
var c = test[curves];
var s = "{{";
for (var i = 0; i < c.length; i += 2) {
s += "{";
s += c[i] + "," + c[i + 1];
s += "}";
if (i + 2 < c.length) {
s += ", ";
}
}
console.log(s + "}},");
}
console.log("</div>");
break;
case 'm':
draw_midpoint = (draw_midpoint + 1) % 3;
redraw();
break;
case 'N':
testIndex += 9;
case 'n':
testIndex = (testIndex + 1) % tests.length;
drawTop();
break;
case 'o':
draw_order ^= true;
redraw();
break;
case 'P':
testIndex -= 9;
case 'p':
if (--testIndex < 0)
testIndex = tests.length - 1;
drawTop();
break;
case 'q':
draw_quarterpoint = (draw_quarterpoint + 1) % 3;
redraw();
break;
case 'r':
for (var i = 0; i < testDivs.length; ++i) {
var title = testDivs[i].id.toString();
if (title == testTitles[testIndex]) {
var str = testDivs[i].firstChild.data;
parse(str, title);
var original = tests.pop();
testTitles.pop();
tests[testIndex] = original;
break;
}
}
redraw();
break;
case 's':
draw_sortpoint = (draw_sortpoint + 1) % 3;
redraw();
break;
case 't':
draw_t ^= true;
redraw();
break;
case 'u':
draw_closest_t ^= true;
redraw();
break;
case 'v':
draw_tangents = (draw_tangents + 1) % 4;
redraw();
break;
case 'x':
draw_point_xy ^= true;
redraw();
break;
case 'y':
draw_mouse_xy ^= true;
redraw();
break;
case '\\':
retina_scale ^= true;
drawTop();
break;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
case '`':
++focus_on_selection;
if (focus_on_selection >= tests[testIndex].length) {
focus_on_selection = 0;
}
setScale(xmin, xmax, ymin, ymax);
redraw();
break;
case '.':
draw_id ^= true;
redraw();
break;
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
}
}
function doKeyDown(evt) {
var char = evt.keyCode;
var preventDefault = false;
switch (char) {
case 37: // left arrow
if (evt.shiftKey) {
testIndex -= 9;
}
if (--testIndex < 0)
testIndex = tests.length - 1;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (evt.ctrlKey) {
redraw();
} else {
drawTop();
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
preventDefault = true;
break;
case 39: // right arrow
if (evt.shiftKey) {
testIndex += 9;
}
if (++testIndex >= tests.length)
testIndex = 0;
pathops version two R=reed@google.com marked 'no commit' to attempt to get trybots to run TBR=reed@google.com Review URL: https://codereview.chromium.org/1002693002
2015-03-24 14:28:17 +00:00
if (evt.ctrlKey) {
redraw();
} else {
drawTop();
}
Mike R: please sanity check SkPostConfig.h Mike K: please sanity check Test.cpp and skia_test.cpp Feel free to look at the rest, but I don't expect any in depth review of path ops innards. Path Ops first iteration used QuickSort to order segments radiating from an intersection to compute the winding rule. This revision uses a circular sort instead. Breaking out the circular sort into its own long-lived structure (SkOpAngle) allows doing less work and provides a home for caching additional sorting data. The circle sort is more stable than the former sort, has a robust ordering and fewer exceptions. It finds unsortable ordering less often. It is less reliant on the initial curve tangent, using convex hulls instead whenever it can. Additional debug validation makes sure that the computed structures are self-consistent. A new visualization tool helps verify that the angle ordering is correct. The 70+M tests pass with this change on Windows, Mac, Linux 32 and Linux 64 in debug and release. R=mtklein@google.com, reed@google.com Author: caryclark@google.com Review URL: https://codereview.chromium.org/131103009 git-svn-id: http://skia.googlecode.com/svn/trunk@14183 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-14 17:08:59 +00:00
preventDefault = true;
break;
}
if (preventDefault) {
evt.preventDefault();
return false;
}
return true;
}
function calcXY() {
var e = window.event;
var tgt = e.target || e.srcElement;
var left = tgt.offsetLeft;
var top = tgt.offsetTop;
mouseX = (e.clientX - left) / scale + srcLeft;
mouseY = (e.clientY - top) / scale + srcTop;
}
function calcLeftTop() {
srcLeft = mouseX - screenWidth / 2 / scale;
srcTop = mouseY - screenHeight / 2 / scale;
}
function handleMouseClick() {
if (!draw_t || !ptInTControl()) {
calcXY();
} else {
redraw();
}
}
function initDown() {
var test = tests[testIndex];
var bestDistance = 1000000;
activePt = -1;
for (var curves in test) {
var testCurve = test[curves];
if (testCurve.length != 4 && testCurve.length != 6 && testCurve.length != 8) {
continue;
}
for (var i = 0; i < testCurve.length; i += 2) {
var testX = testCurve[i];
var testY = testCurve[i + 1];
var dx = testX - mouseX;
var dy = testY - mouseY;
var dist = dx * dx + dy * dy;
if (dist > bestDistance) {
continue;
}
activeCurve = testCurve;
activePt = i;
bestDistance = dist;
}
}
if (activePt >= 0) {
lastX = mouseX;
lastY = mouseY;
}
}
function handleMouseOver() {
calcXY();
if (draw_mouse_xy) {
var num = mouseX.toFixed(decimal_places) + ", " + mouseY.toFixed(decimal_places);
ctx.beginPath();
ctx.rect(300, 100, num.length * 6, 10);
ctx.fillStyle = "white";
ctx.fill();
ctx.font = "normal 10px Arial";
ctx.fillStyle = "black";
ctx.textAlign = "left";
ctx.fillText(num, 300, 108);
}
if (!mouseDown) {
activePt = -1;
return;
}
if (activePt < 0) {
initDown();
return;
}
var deltaX = mouseX - lastX;
var deltaY = mouseY - lastY;
lastX = mouseX;
lastY = mouseY;
if (activePt == 0) {
var test = tests[testIndex];
for (var curves in test) {
var testCurve = test[curves];
testCurve[0] += deltaX;
testCurve[1] += deltaY;
}
} else {
activeCurve[activePt] += deltaX;
activeCurve[activePt + 1] += deltaY;
}
redraw();
}
function start() {
for (var i = 0; i < testDivs.length; ++i) {
var title = testDivs[i].id.toString();
var str = testDivs[i].firstChild.data;
parse(str, title);
}
drawTop();
window.addEventListener('keypress', doKeyPress, true);
window.addEventListener('keydown', doKeyDown, true);
window.onresize = function () {
drawTop();
}
}
</script>
</head>
<body onLoad="start();">
<canvas id="canvas" width="750" height="500"
onmousedown="mouseDown = true"
onmouseup="mouseDown = false"
onmousemove="handleMouseOver()"
onclick="handleMouseClick()"
></canvas >
</body>
</html>
Reference in New Issue Copy Permalink