Age | Commit message (Collapse) | Author | |
---|---|---|---|
2014-01-07 | clean up test runner output | Alon Zakai | |
2014-01-06 | benchmark updates | Alon Zakai | |
2014-01-06 | refactor lua benchmark | Alon Zakai | |
2014-01-06 | option to benchmark multiple llvm versions, both native and js | Alon Zakai | |
2014-01-06 | other.test_emcc fixes for fastcomp | Alon Zakai | |
2014-01-04 | fix function type in test_openal_buffers for asm | Alon Zakai | |
2014-01-04 | specify (void) as the arguments in webgl_context_attributes test, to avoid ↵ | Alon Zakai | |
confusion with varargs by clang | |||
2014-01-03 | disable/update various browser tests for fastcomp | Alon Zakai | |
2014-01-03 | fix benchmark suite bug where args got appended | Alon Zakai | |
2014-01-03 | disable named globals test in fastcomp | Alon Zakai | |
2014-01-03 | convert test_sdl_canvas to btest | Alon Zakai | |
2014-01-03 | [SDL] Fixing SDL_UnlockSurface in IE10/IE11. | John Vilk | |
Previously, just calling SDL_UnlockSurface in IE10/IE11 would throw an exception, since Emscripten assumed that the ImageData's `data` property was Uint8ClampedArray, which has a backing buffer. IE10/IE11 still uses the deprecated CanvasPixelArray, which does not have a backing buffer property: https://developer.mozilla.org/en-US/docs/Web/API/CanvasPixelArray I've added an additional path to SDL_UnlockSurface for these browsers. | |||
2014-01-03 | disable fc in default benchmarks | Alon Zakai | |
2014-01-03 | fix output in glGetFramebufferAttachmentParameteriv | Alon Zakai | |
2014-01-03 | properly handle getParameter returning null when it means a null object ↵ | Alon Zakai | |
(e.g., no buffer is bound) and not an invalid name | |||
2014-01-03 | disable test_emscripten_log in spidermonkey due to breakage; issue #1970 | Alon Zakai | |
2013-12-31 | sanity check for js backend in fastcomp | Alon Zakai | |
2013-12-30 | disable some other.* tests in fastcomp | Alon Zakai | |
2013-12-30 | fix --save-bc in fastcomp | Alon Zakai | |
2013-12-30 | fix other.test_crunch | Alon Zakai | |
2013-12-30 | Merge pull request #1946 from juj/default_obj_ext | juj | |
--default-obj-ext | |||
2013-12-29 | Define behavior that 'emcc -c a.c -o dir/' shall compile and generate object ↵ | Jukka Jylänki | |
file 'dir/a.o'. Previous behavior when directory was specified in -o was accidental, and it generated an object file 'dir/a_0.o'. Add new command line option --default-obj-ext that allows specifying the file suffix that is used when the output object filename is generated in this manner. '.o' is a good default suffix since it parallels the existing gcc/clang/linux convention. For Windows Visual Studio+CMake+Emscripten integration, CMake has been hardcoded to assume that if Visual Studio is targeted, the compiler will always generate '.obj' files. Hence having the ability to adjust the default naming scheme with --default-obj-ext enables working around CMake inflexibility, and add support for CMake+VS+Emscripten triple. | |||
2013-12-26 | fix js libraries in fastcomp; fixes #1948 | Alon Zakai | |
2013-12-25 | disable new test in fastcomp | Alon Zakai | |
2013-12-25 | properly handle invoke of setjmp, by lowering to a call, which fixes phis ↵ | Alon Zakai | |
that refer to it. fixes #1942 | |||
2013-12-24 | disable a test requiring unaligned memory in fastcomp | Alon Zakai | |
2013-12-24 | properly check for libx.so.1.2.3 types when deciding what to link in ↵ | Alon Zakai | |
statically, and improve testing for that; fixes #1886 | |||
2013-12-23 | do not link in dynamic libraries when compiling to bitcode in intermediate ↵ | Alon Zakai | |
stages, only link them in when building to js or html. closes #1886 | |||
2013-12-23 | fix bug where close-together tempDoublePtr operations could cross each other | Alon Zakai | |
2013-12-23 | improve fuzz test tool | Alon Zakai | |
2013-12-22 | do not use named globals in fastcomp | Alon Zakai | |
2013-12-21 | save each benchmark to a unique name | Alon Zakai | |
2013-12-21 | benchmark fastcomp | Alon Zakai | |
2013-12-21 | Update test other.test_demangle to pass after the previous commit which ↵ | Jukka Jylänki | |
added demangling '_main' to 'main()'. | |||
2013-12-20 | update browser.test_preload_file | Alon Zakai | |
2013-12-20 | update tests for recent file packager changes | Alon Zakai | |
2013-12-20 | fuzz i64 math as well | Alon Zakai | |
2013-12-20 | do not force asm in fuzzer | Alon Zakai | |
2013-12-20 | randomize opt level in csmith driver | Alon Zakai | |
2013-12-20 | fuzz testcase | Alon Zakai | |
2013-12-20 | Merge pull request #1922 from coolwanglu/embed_dot_files | Alon Zakai | |
--exclude-file | |||
2013-12-20 | Merge pull request #1635 from juj/emscripten_log | Alon Zakai | |
emscripten_log() and emscripten_get_callstack() | |||
2013-12-20 | fuzz testcase | Alon Zakai | |
2013-12-20 | fuzz testcase | Alon Zakai | |
2013-12-20 | improve fuzz testcase runner | Alon Zakai | |
2013-12-20 | simplify ir for fastcomp after the autodebugger runs | Alon Zakai | |
2013-12-20 | make fuzz tester executable | Alon Zakai | |
2013-12-20 | csmith fixes | Alon Zakai | |
2013-12-20 | Fix browser.test_emrun to utilize the EMSCRIPTEN_BROWSER env. variable like ↵ | Jukka Jylänki | |
the rest of the browser suite does, instead of always testing user default browser. | |||
2013-12-20 | In browser.test_emrun, use --log_stdout and --log_stderr to capture run ↵ | Jukka Jylänki | |
output instead of python pipe, since on linux, spawning chrome browser would run into a mysterious issue where piping the emrun output would hang the python subprocess spawn. |