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
|
2011-06-22: Version 1.3
* Optional typed arrays with single shared buffer (TA2)
* Optional support for nonportable optimizations with TA2
* Relooper optimizations
* Reading from stdin on the web opens window.prompt
* Various bug fixes
2011-05-29: Version 1.2
* Doom demo
* Major SDL improvements: color palettes, input events, audio
* Various improvements for CHECK_* and CORRECT_*, and new AUTO_OPTIMIZE
* Experimental work towards supporting OpenGL in WebGL
* Various bug fixes
2011-05-01: Version 1.1
* Much improved Bullet demo (llvm opts, quantum = 1, CubicVR.js)
* QUANTUM == 1 a.k.a memory compression option
* Dead function elimination tool
* Various performance improvements in generated code
* Various bug fixes
2011-04-09: Version 1.0
* Poppler test and web demo
* Optimize compiler memory usage very significantly
* Support for LLVM 2.9
* Better interaction with closure compiler
* Finish docs/paper.pdf
* Many bug fixes
2011-03-05: Version 0.9
* OpenJPEG test and web demo
* Line number debugging info and autodebugger tool
* CORRECT_ROUNDINGS option
* Line-specific CORRECT_* options
* 20% faster compilation
* Generate strict mode JavaScript
* Many bug fixes, additional tests and library implementations
2011-02-06: Version 0.8
* Freetype web demo (2011-02-08, right after 0.8)
* Freetype and zlib tests (including the entire build procedure)
* File emulation
* CHECK_OVERFLOWS & CORRECT_OVERFLOWS options to handle numerical size issues
* CHECK_SIGNS option to find whether GUARD_SIGNS is necessary
* Improvements to usage of llvm optimizations, and testing
* Many bug fixes, additional tests and library implementations
2010-12-18: Version 0.7
* Python web demo
* Support for essentially all LLVM optimizations
* Proper support for C bitfields
* Many bug fixes, additional tests and library implementations
2010-11-25: Version 0.6
* Web demos: Lua and Bullet/WebGL
* SAFE_HEAP checks for invalid reads/writes, nonportable .ll
* Basic C++ exceptions support (|catch(...)|)
* Optimize compilation of very large projects (memory and speed)
* Support for frontend-optimized .ll input, plus tests
* Integration (combining scripts with compiled C++) tools and tests
* Many bug fixes, additional tests and library implementations
2010-11-02: Version 0.5
* Much faster compilation, in particular of large projects
* Bullet physics library test
* GCC name demangling test
* Module-ization of generated code (optional)
* Name demangler and namespace generator tools
* Many code cleanups, bug fixes, additional tests and library implementations
2010-10-17: Version 0.4
* Much faster optimized code, now 10X the speed of our unoptimized code
* Support for the recently-released LLVM 2.8
* Support for typed arrays
* Integration with the Closure Compiler
* Benchmarking framework in test runner
* Many code cleanups, bug fixes, and additional tests
2010-10-05: Version 0.3
* Much faster compilation (but still slow with relooper)
* Clang support
* Optional memory alignment that precisely matches C/C++
* Proper memory management, including stack and (optional) dlmalloc
* Rewritten relooper; no emulated blocks in any test
* Initial support for SDL
* Raytracing test + web demo
* Many code cleanups, bug fixes, additional tests and library implementations
2010-09-11: Version 0.2
* sauer (cubescript) test passes (without RELOOPing), + web demo
* ES_SIZEOF - safe&portable sizeof replacement
* emscripten.py tool for easy compiling
* Better debugging support, using SAFE_HEAP and LABEL_DEBUG, using internal preprocessor
* Compiler can now run all tests in both SpiderMonkey and V8
* Various compiler optimizations (still barely scratched the surface though)
* Many code cleanups, bug fixes, additional tests and library implementations
2010-08-28: Version 0.1
* All tests pass, including fannkuch and fasta, but constglobalstructs
* Relooping of Fannkuch is complete, fasta has one left
* Emscriptened Fannkuch is 19X slower than gcc -O0, 37X than gcc -O2
|
