summaryrefslogtreecommitdiff
path: root/emcc
AgeCommit message (Collapse)Author
2013-02-26fix DEBUG ? 2 checksAlon Zakai
2013-02-26add cache debuggingVladimir Vukicevic
2013-02-26clean up response file codeAlon Zakai
2013-02-26Add basic response file parsing to emccVladimir Vukicevic
2013-02-25EMCC_FORCE_STDLIBS optionAlon Zakai
2013-02-25only do symbol checks for libcxx etc. if not forcing them anyhowAlon Zakai
2013-02-21Merge pull request #859 from waywardmonkeys/cmdline-fixesAlon Zakai
Cmdline fixes
2013-02-21emcc: Improve handling of -O flag.Bruce Mitchener
Previously, while gcc supports -O, emcc would fail. It now interprets -O as -O2, just as gcc does. Additionally, it was only looking at the first character after -O, so nonsense like -O34 would be accepted (and treated as -O3).
2013-02-20do not wipe out llvm debug info if linespecific corrections are doneAlon Zakai
2013-02-20remove bad_cast/typeinfo.cpp from libcxx, no longer needed since it is in ↵Alon Zakai
libcxxabi
2013-02-20build base type_info class with libcxxabi, fixes #862Alon Zakai
2013-02-19closure docs on emcc --helpAlon Zakai
2013-02-18run closure in -O3, it makes sense there with the other unsafe optimizationsAlon Zakai
2013-02-18remove LLVM debug info in -O1+, since the optimizer removes it anyhowAlon Zakai
2013-02-18fix emcc --version in other dirsAlon Zakai
2013-02-17default closure to off in all optimization levels - we rely on registerize ↵Alon Zakai
for much faster effects on runtime speed
2013-02-17use registerize only when not keeping debug, and also use it when closure is ↵Alon Zakai
disabled, not just ASM_JS mode (which is just one case where closure is disabled)
2013-02-17disable closure if debug info was requested (-g)Alon Zakai
2013-02-17show version and git revision in emcc --versionAlon Zakai
2013-02-03Merge pull request #825 from juj/stdAlon Zakai
Std
2013-02-03do not run unnecessary optimizeShiftsAggressive+eliminate steps in asm mode, ↵Alon Zakai
it is pointless and just wastes compilation time
2013-02-02do not build musl memcpyAlon Zakai
2013-02-01docs for file packagerAlon Zakai
2013-02-01Compile with -std=c++03 consistently on all platforms. By default, Clang 3.2 ↵Jukka Jylänki
inconsistently uses C++11 for Windows and C++03 for Linux if user did not specify -std= on the command line. Fixes other.test_embind and *.test_llvmswitch. Add unit tests that confirm that .cpp files are compiled with C++03 std by default.
2013-01-31hand-optimize memcpy/asmAlon Zakai
2013-01-19remove emcc asm test wrappingAlon Zakai
2013-01-13do not run js optimizer needlessly on 'last'Alon Zakai
2013-01-12emit 0.0 in asm double definitionsAlon Zakai
2013-01-12fix emcc calls to js optimizerAlon Zakai
2013-01-12refactor use of asm in js optimizerAlon Zakai
2013-01-10export memcpy in asm mode so non-asm library functions can use itAlon Zakai
2013-01-10fix compiling of postrinsic memcpy when in forced c++ mode for the main fileAlon Zakai
2013-01-10link in memcpy after opts and lto if it shows up only at that timeAlon Zakai
2013-01-09use musl memcpy instead of out handwritten js oneAlon Zakai
2013-01-09rename dlmalloc to libc in cache, in preparation for adding further code there1.2.3Alon Zakai
2013-01-08make docs mention that bitcode is the defaultAlon Zakai
2013-01-08keep a coercion right on top of heap accesses in asm modeAlon Zakai
2013-01-07TODO for asm double literalsAlon Zakai
2013-01-03warn about ASM_JSAlon Zakai
2012-12-31Merge branch 'incoming' into asm_jsAlon Zakai
Conflicts: src/library_browser.js
2012-12-24move llvm opt debug messagesAlon Zakai
2012-12-21merge incomingAlon Zakai
2012-12-21improve ltoAlon Zakai
2012-12-21avoid separate lto passAlon Zakai
2012-12-21safely internalize in s_x_x tests tooAlon Zakai
2012-12-21do lto before dce, since lto opens up dce opportunitiesAlon Zakai
2012-12-21be careful with internalize in llvm LTO, and mark needed symbols for relooperAlon Zakai
2012-12-21llvm 3.2 internalize internalizes even main, so manually keep it and other ↵Alon Zakai
exports alive
2012-12-19Updated to add a PYTHON config option instead of defaulting to python2. This ↵Alan Kligman
should preserve the default behavior in a way that can be overridden.
2012-12-19Updated scripts to call python2 directly rather than relying on python ↵Alan Kligman
symlink pointing to the right place. See PEP394 for details on why this should be OK.