emscripten/system/include, branch 1.7.2 Emscripten: An LLVM-to-JavaScript Compiler https://git.amat.us/emscripten/atom/system/include?h=1.7.2 2013-11-07T09:08:02Z Don't use atomic as that breaks our version of clang. 2013-11-07T09:08:02Z Bruce Mitchener bruce.mitchener@gmail.com 2013-11-07T09:08:02Z urn:sha1:8a8f0348a15b274ab2fc1f98a434fe8c7e02ff82 Update libcxx to 194185, 2013-11-07. 2013-11-07T09:07:18Z Bruce Mitchener bruce.mitchener@gmail.com 2013-11-07T08:49:37Z urn:sha1:44af976a44bc194b985fdb880f99a7feff133b5a This brings C++14 support. Merge pull request #1748 from caiiiycuk/fix_sdl_colors 2013-11-06T19:56:14Z Alon Zakai alonzakai@gmail.com 2013-11-06T19:56:14Z urn:sha1:7f870cf9c357f6a1138ba612ace7d7249f85e250 Fix SDL_SetColors && SDL_envets begin to add emscripten_float32x4_signmask 2013-10-30T20:14:26Z Alon Zakai alonzakai@gmail.com 2013-10-30T20:14:26Z urn:sha1:a4b5ac5a0b7b705d89def4fb447f36cbe04d5789 Fix SDL_SetColors && SDL_envets 2013-10-30T11:18:29Z Aleksander Guryanov caiiiycuk@gmail.com 2013-10-30T11:18:29Z urn:sha1:2247ff845fa862a35fc9d1acf8150eb7e56f41ac Fix an issue with Boost compilation. 2013-10-27T06:14:22Z Bruce Mitchener bruce.mitchener@gmail.com 2013-10-27T06:02:37Z urn:sha1:5017df14dbeadcb486a5b8b95533142ee4296fd1 The issue is that when these macros are defined, libcxx creates inline functions and so we end up with 2 separate defintions of these functions (one inline in std and one that is extern "C"). We undef these until libcxx is fixed. Without this, some things can fail to compile correctly, like Boost. Fixes issue #1716. add vector.h 2013-10-18T20:00:49Z Alon Zakai alonzakai@gmail.com 2013-10-18T18:01:53Z urn:sha1:634513d0c75e7d09a9c9055d4077d631e443ecb4 make EM_ASM use strings on the heap, so it works in asm, and cache the generated functions 2013-10-07T00:44:06Z Alon Zakai alonzakai@gmail.com 2013-10-07T00:44:06Z urn:sha1:332325394c20bb4a3972344144803fcd93476b08 Mark __assert_fail as _Noreturn. 2013-09-24T14:01:04Z Bruce Mitchener bruce.mitchener@gmail.com 2013-09-24T14:01:04Z urn:sha1:baa4d38f13f436bdeba724f0825bbe985637be22 This eliminates a performance regression caused in the box2d benchmark by switching to the musl headers. Without this being _Noreturn, clang / llvm generates much worse code involving phis and extra loads. emscripten_async_load_script 2013-09-13T22:13:30Z Alon Zakai alonzakai@gmail.com 2013-09-13T22:13:30Z urn:sha1:1fc6762e3176843ffc378c8b4f43dd73fed25ebe