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