-
Notifications
You must be signed in to change notification settings - Fork 13
/
ChangeLog
192 lines (168 loc) · 6.46 KB
/
ChangeLog
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
CHANGES
=======
1.6.0
-----
* release(1.6.0): faster 6 and 18-way dijkstra
* fix: CC -> CXX
* chore: add py39 to build
* perf: improve speed for dijkstra 6-way and 18-way
* chore(build): add infrastructure for building manylinux2010 & 2014
1.5.1
-----
* release(1.5.1): fix broken py38 release for MacOS
* chore: update changelog
1.5.0
-----
* release(1.5.0): adds free\_space\_radius to EDF
* release(1.5.0): adds free\_space\_radius to EDF
* chore: more requirements\_dev.txt changes
* chore: create requirements\_dev.txt
* chore: add tox.ini for mac builds
* chore: add appveyor.yml for windows builds
* chore: update to production/stable and update python versions to tested ones
* test: add 3.8 to travis
* perf: euclidean distance field acceleration (#11)
1.4.2
-----
* release(1.4.2): fix 64-bit addressable in euclidean distance field
* fix: ensure index is large enough for large images
1.4.1
-----
* release(1.4.1): update libdivide to support clang compilation
* chore: update libdivide.h
* chore: update Changelog
1.4.0
-----
* release(1.4.0): support 64-bit addressable arrays
* feat: support arrays with 64-bit indices (#10)
* docs: link connectivity numbers to their respective distance metrics
* docs: eliminate duplicate text, specify finite weights
1.3.1
-----
* release(1.3.1): improved heuristics for 6 and 18 connected compass
* perf: add decent heuristic for 18 connected
* fix: don't modify a const variable
* chore: update ChangeLog
* perf: use correct distance measure for 6 connected
* fix: don't modify a const variable
* chore: update ChangeLog
1.3.0
-----
* release(1.3.0): adds connectivity and compass parameters
* docs: improve performance test and documentation
* docs: add random start/target performance numbers
* docs: add description of how to use cpp compass search
* docs: update benchmark with A\*
* docs: show how to use the compass feature
* test: additional tests for A\* search
* feat: add A\* search via "compass" parameter to djikstra (#6)
* feat: accept 4 and 8 connectivity for 2D images
* fix: update description of connectivity for 2D
* docs: add connectivity to README.md
* feat: add support for 6 and 18 connectivities (#8)
* chore: update ChangeLog
1.2.0
-----
* release(1.2.0): bidirectional search option
* docs: update with new benchmark
* docs: add reference to original bidirectional search paper
* feat: bidirectional dijkstra (#5)
* docs: reword some sentences in README.md
* fix: make python3.5 compile better on macOS
* fix: replace size\_t with uint64\_t to make libdivide work on mac
1.1.0
-----
* release(1.1.0): fortran + removing copy ops
* feat+perf: mandate fortran order arrays + avoid copies (#4)
* refactor: not necessary to pass loc to compute\_neighborhood
* test: source == target
* perf: simplify search when source == target
* chore: update setup.cfg with markdown description content type
* Update README.md
* docs: added reference for early termination technique
* chore: drop py34 from travis testing
1.0.1
-----
* chore: drop py34 binary support
* chore: version 1.0.1
* refactor: remove experimental code comment
* perf: use libdivide.h for faster x y z computation
* docs: added link to PyPI
1.0.0
-----
* chore: added py37 to setup.cfg
* test: add Python 3.7 to travis testing
* chore: add documentation and rename dijkstra to dijkstra3d
* chore: setting up for PyPI distribution
* test: fill out cpp test code with an example
* docs: update timestamp
* fix: euclidean\_distance\_field could be succeptible to non-boolean fields
* perf: faster dijkstra3d
* perf: faster euclidean distance field
* perf: 10% faster
* perf: bug in distance\_field3d was hurting performance a lot
* fix: rows/cols swapped in path\_from\_parents
* docs: updated module help
* perf: added parental\_field and path\_from\_parents
* docs: added anisotropy to euclidean\_distance\_field help
* feat: euclidean distance field
* feat: supported uints more fully for dijkstra
* doc: why does fortran order work? idk
* fix: simplified distance\_field
* fix: reconciled various transpositions
* test: added more complex test to dijkstra
* fix: algorithm was not properly checking visited
* fix: handle fortran order (by converting to C order..)
* docs: elaborated on where different contributions to memory usage come from
* docs: added benchmark
* feat: added support for uints and bools
* docs: added reference to Dijkstra paper
* docs: how to use distance\_field, performance info
* docs: added travis badge
* test: added 3d tests
* test: added automated tests
* fix: pyx handle Fortran ordered arrays in distance\_field
* fix: memory leak in distance field calculation
* fix: accomodate 2d and 3d inputs to dijkstra
* fix: row col depth assignment in distance\_field
* fix: corrected neighbor calculation and signed bit is actually needed
* feat: added prototype of distance\_field function
* docs: show how to use
* fix: made point cloud output sensible for 2D
* refactor: broke dijkstra function into components for neatness
* feat: running python code
* feat: Cython bindings for dijkstra3d
* feat: added dijkstra namespace, 2d and 3d functions
* refactor: make x,y,z translation more explicit
* perf: mostly eliminated compute\_neighborhood as a CPU suck
* perf: shaved another 10sec off the 512^3
* fix: removed pairing\_heap.hpp include
* perf: switched to std::priority\_queue, 20% speedup
* feat: moving to 26-hood
* feat: moved to 18-hood from 6-hood
* refactor: allocate heap instance on stack (not the full tree)
* refactor: allocate neighborhood on the stack
* fix: y coordinate computed incorrectly
* feat: now returns st path as a vector of array indicies
* perf: break out of the loop early, saves huge computation
* perf: a more dangerous, but highly performant improvement to running time
* perf: faster delmin through reserving vector size
* fix: dijkstra working but kinda slow, but very low memory
* fix: compilation warnings
* fix: improper deletion of heap
* fix: segmentation fault from accessing dist after free
* wip: moving towards a heap based dijkstra
* chore: add -std=c++11 to test compilation
* perf: why use two operations when you can have one
* fix: don't segfault when printing with no root
* perf: ensure compiler can optimize forest.size()
* fix: propogated "last" fix for odd sizes
* fix: pairing heap now sorts numbers
* docs: fixed formatting in a comment
* test: added heap sort test
* fix: construct PHNode correctly
* fix: syntax errors in pairing\_heap.hpp
* feat: untested pairing heap implementation
* docs: added paper references
* feat: a classic algorithm, O(V^2)
* Initial commit