emscripten - Emscripten: An LLVM-to-JavaScript Compiler

Age	Commit message (Collapse)	Author
2013-12-01	fix for pull #1882	Alon Zakai

2013-12-01	differentiate external vars from funcs in fastcomp	Alon Zakai

2013-12-01	Merge pull request #1882 from abergmeier/safe_ensure_dirs	Alon Zakai
	Introduce safe_ensure_dirs as a safe os.makedirs replacement.
2013-12-01	Introduce safe_ensure_dirs as a safe os.makedirs replacement.	Andreas Bergmeier
	It only uses os.makedirs call, which is more atomic than calling 2 procs. Thus exception due to already existent directory should not longer occur.
2013-12-01	fix emscripten_hide_mouse removal of old rule	Alon Zakai

2013-11-30	send EXPORTED_FUNCTIONS to fastcomp glue	Alon Zakai

2013-11-29	fix memmove return value	Alon Zakai

2013-11-29	add missing mkport dependency	Alon Zakai

2013-11-29	emit runPostSets call in fastcomp	Alon Zakai

2013-11-29	fix repeating global_funcs for fastcomp	Alon Zakai

2013-11-29	do not emit empty postsets from js compiler for fastcomp output	Alon Zakai

2013-11-28	fix bitcast type when the input is an llvm function like getelementptr	Alon Zakai

2013-11-28	fix memset return value	Alon Zakai

2013-11-28	fix fastcomp vararg alignment, only 4-byte	Alon Zakai

2013-11-28	Merge pull request #1872 from coolwanglu/mkdtemp	Alon Zakai
	mkdtemp
2013-11-28	Merge branch 'glGet-cleanup' into incoming	Alon Zakai

2013-11-28	cleanup and consolidate glGet	Alon Zakai

2013-11-28	Document GL_FFP_ONLY setting in src/settings.js.	Jukka Jylänki

2013-11-28	check error in mkdtemp()	Lu Wang

2013-11-28	mkdtemp	Lu Wang

2013-11-27	send externs to library including	Alon Zakai

2013-11-27	add dso_handle to library, for fastcomp	Alon Zakai

2013-11-27	add fastcomp externs to asm imports	Alon Zakai

2013-11-27	Merge pull request #1862 from abergmeier/prevent_false_negative_in_cache	Alon Zakai
	Prevent Cache.ensure from throwing an error, while the directory already...
2013-11-27	Merge pull request #1867 from abergmeier/spelling_bee	Alon Zakai
	Fix spelling error.
2013-11-27	Merge pull request #1863 from juj/gl_emulation_opts_rebased	Alon Zakai
	Optimize GL emulation layer.
2013-11-27	GL_IMPLEMENTATION_COLOR_READ_TYPE and GL_IMPLEMENTATION_COLOR_READ_FORMAT	Alon Zakai

2013-11-27	Fix spelling error.	Andreas Bergmeier

2013-11-27	Cast GL emulation matrix version numbers back to integers after increment. ↵	Jukka Jylänki
	Code cleanup and clarify comments.
2013-11-26	tiny code cleanup	Alon Zakai

2013-11-26	Merge pull request #1865 from ngld/llvm-used	Alon Zakai
	Fix @llvm.used
2013-11-27	Set proper value type for @llvm.used	ngld

2013-11-26	Merge pull request #1864 from ngld/llvm-used	Alon Zakai
	Implemented #1573. (@llvm.used)
2013-11-26	Merge pull request #1757 from wahbahdoo/incoming	Alon Zakai
	Incoming -- add support for SIMD x86 intrinsics
2013-11-26	add SIMD contributor to AUTHORS	Heidi Pan

2013-11-26	update simd_test to use updated int32x4 instead of uint32x4; add simd_test3 ↵	Heidi Pan
	for more comprehensive x86 intrinsics unit tests
2013-11-26	getting some initial set of x86 intrinsics mapped to JS SIMD ops; update ↵	Heidi Pan
	existing SIMD code to use updated JS SIMD API (SIMD.type.op instead of SIMD.op & SIMD.opu32; int32x4 instead of uint32x4); bug fix: right type of withX() for int32x4 initialization; bug fix: splat(0) instead of zero() for int32x4 initialization
2013-11-26	getting some initial set of x86 intrinsics mapped to JS SIMD ops; update ↵	Heidi Pan
	existing SIMD code to use updated JS SIMD API (SIMD.type.op instead of SIMD.op & SIMD.opu32; int32x4 instead of uint32x4); bug fix: right type of withX() for int32x4 initialization; bug fix: splat(0) instead of zero() for int32x4 initialization
2013-11-26	Implemented #1573. All functions marked as "used" will be automatically ↵	ngld
	added to EXPORTED_FUNCTIONS.
2013-11-26	Add new GL emulation related #define GL_FFP_ONLY which is used to signal ↵	Jukka Jylänki
	that the C/C++ client code will not use any GL shaders, allowing more efficient optimizations to be performed in the FFP GL emulation code knowing that it doesn't need to worry about shader programs from the user.
2013-11-26	Skip looping twice over textures in GL emulation renderer prepare() code. ↵	Jukka Jylänki
	Clean up unnecessary variable.
2013-11-26	Name GL functions that show up as anonymous in the profile.	Jukka Jylänki

2013-11-26	Version immediate mode GL matrix stack data, so that the matrices don't need ↵	Jukka Jylänki
	to be reset on each program if they didn't change.
2013-11-26	Fix GL logging message when GL_UNSAFE_OPTS is disabled.	Jukka Jylänki

2013-11-26	Cache the currently active FFP emulation renderer object. If ↵	Jukka Jylänki
	renderer-related GL state does not change, the renderer can then be looked up very quickly, instead of having to recompute the key every time.
2013-11-26	Optimize GL emulation performance by packing the GLTexEnv renderer key to a ↵	Jukka Jylänki
	tighter space, which greatly reduces the number of required CKeyView.next() traversals that show up costly in the profiles.
2013-11-26	Simplify the key state used for identifying used vertex attribute set. The ↵	Jukka Jylänki
	function CKeyView.next() shows up as one of the hottest in profiles, so merging multiple key identifiers to one helps performance.
2013-11-26	Remove vertex attribute size and type information from being part of the FFP ↵	Jukka Jylänki
	emulation renderer key. This improves performance by making the renderer key simpler, which causes fewer renderer state key traversals to be done in getRenderer(), and by reducing the number of created shader programs and # of shader program switches performed during frame.
2013-11-26	Added new configuration parameter support 'GL_MAX_TEXTURE_IMAGE_UNITS' to ↵	Jukka Jylänki
	the emscripten Module object so that codebases using fixed-function GL emulation mode can hint to the emulation code how many texture units they are using at maximum. This optimizes the emulation code path to avoid having to examine an unnecessary amount of texture units at each draw call (at minimum the code would loop over to confirm that the unused texunits are really unused, which would burn up some unneeded cycles.)
2013-11-26	parse hex floats in vector literals	Alon Zakai