Skip to content

Conversation

@Cellule
Copy link
Contributor

@Cellule Cellule commented Jul 12, 2016

Integration latest master to WebAssembly
Remove DoubleExtended OpCode implementation
Update host to use newer Jsrt utf8 apis


This change is Reviewable

dilijev and others added 30 commits May 16, 2016 13:20
Change all file names (files/baseline) in rlexe.xml to match real file
names on disk.
Prevent mixed output: Previously a test run sends/queues multiple result
messages. This can result in intertwined messages from different test runs
thus mixed output. Change test run _print to buffer outputs, and sends
whole test pass/fail/elapsed_time together with output as one message.

Add `exclude_x86/x64' to global `not_tags` to better filter tests.

Support multiple `tags` in rlexe.xml. Used by some tests.

Support test specific `timeout`.

Changed log format slightly to log [#current_test/#total_test] to indicate
progress.
Merge pull request chakra-core#1009 from jianchun:rlxmlcase
Change all file names (files/baseline) in rlexe.xml to match real file
names on disk.
…sc improvements

Merge pull request chakra-core#1010 from jianchun:py2
Prevent mixed output: Previously a test run sends/queues multiple result
messages. This can result in intertwined messages from different test runs
thus mixed output. Change test run _print to buffer outputs, and sends
whole test pass/fail/elapsed_time together with output as one message.

Add `exclude_x86/x64' to global `not_tags` to better filter tests.

Support multiple `tags` in rlexe.xml. Used by some tests.

Support test specific `timeout`.

Changed log format slightly to log [#current_test/#total_test] to indicate
progress.
`wcscmp` produces positive or negative results based on the inputs. However we expect -1, 0, +1

Also, normalized another usage of `wcscmp` although no consumer found for `Compare`.
This issue was preventing WScript.Echo or print functions to output a unicode string.
While running tests we are more interested in failed tests. Printing
passed results on the screen makes it harder to read failed tests.

Change `runtests.py` slightly so that "Pass" tests are still printed on the
terminal, but immediately overwritten by the next result. Now we can still
read test progress but passed tests no longer flush the screen.
Exclude more Simd tests (require_backend), Debugger tests (
require_debugger), and apply exclude_chk/exclude_fre.

Exclude tests that require compile-flags: -serialized, -simdjs (neither
implemented yet on cross-platform).
Merge pull request chakra-core#1024 from jianchun:py
runtests.py: "hide" passed tests on the terminal

```
While running tests we are more interested in failed tests. Printing
passed results on the screen makes it harder to read failed tests.

Change `runtests.py` slightly so that "Pass" tests are still printed on the
terminal, but immediately overwritten by the next result. Now we can still
read test progress but passed tests no longer flush the screen.
```

xplat: exclude more unsupported tests

```
Exclude more Simd tests (require_backend), Debugger tests (
require_debugger), and apply exclude_chk/exclude_fre.

Exclude tests that require compile-flags: -serialized, -simdjs (neither
implemented yet on cross-platform).
```
Merge pull request chakra-core#1025 from obastemur:obastemur_fix_fwrite
Fixes unicode output to file stream.

```
print("ab\u00A0cd");

before fix : ab c
after fix  : ab cd
```
- Turn off fpo- stack walking depends on it on retail builds
- Speed up debug builds by improving performance of IsAddressOnStack
- Add support for debug and retail builds being in different directories
Also, updated var name based on CR feedback
Merge pull request chakra-core#1032 from digitalinfinity:linux_ss
- Turn off fpo- stack walking depends on it on retail builds
- Speed up debug builds by improving performance of IsAddressOnStack
- Add support for debug and retail builds being in different directories
…o build using clang on windows

Merge pull request chakra-core#949 from curtisman:clang
Build x64 no jit configurations using clang on windows.
Disable a bunch of warning,
Fix issue with PCH builds.

Since clang on windows doesn't support /homeparam, any javascript that requires stack walking doesn't work (e.g. exception)
Merge pull request chakra-core#1021 from obastemur:obastemur_fix_wcscmp
`wcscmp` produces positive or negative results based on the inputs. However we expect -1, 0, +1

Also, normalized another usage of `wcscmp` although no consumer found for `Compare`.
 - fix -> on some platforms `pow` doesn't support `-inf`
 - add test case for non-integer negative exponent for -0
 - add test case for negative integer exponent for +0
 - remove PAL::pow
…ything lines up better and they are easier to discover.

Merge pull request chakra-core#993 from dilijev:build
Merge pull request chakra-core#1018 from obastemur:fix_pow
 - fix -> on some platforms `pow` doesn't support `-inf`
 - add test case for non-integer negative exponent for -0
 - add test case for negative integer exponent for +0
…ith JIT enabled.

Merge pull request chakra-core#1043 from curtisman:nosimd
- Implements DateTime support Cross-Platform
- Redesigns DaylighTime related API
Merge pull request chakra-core#975 from obastemur:date_work
- Implements DateTime support Cross-Platform
- Redesigns DaylightHelper and Date utilities
- Replaces chakra-core#915
…d force build flavor

Merge pull request chakra-core#1040 from obastemur:obastemur_fix_runtests
Similar to Windows runtests.cmd, runtests.py asks for a build flavor instead of forcing release build.
issue: previously implemented name caching was adding an additional character to
the end of display name.

extra: consumer for GetStandardName and GetDaylightName asks only for one of them
while the previous implementation was copying both. Keep timeZoneInfo as
is and bring the result back from that.
akroshg and others added 21 commits July 7, 2016 18:53
…ault on.

Merge pull request chakra-core#1257 from akroshg:stringpad

Apart from enabling it, I found that spec for padStart/padEnd got changed where ,
     8. If filler is the empty String, return S. (https://tc39.github.io/ecma262/#sec-string.prototype.padstart)
fixed that and updated the test.
test262 is fully passed for these two APIs.
Added a check to return value of fread. Updated the copy-right. Added jenkins support.
Merge pull request chakra-core#1256 from akroshg:nta1

Added a check to return value of fread.  Added jenkins support.
Merge pull request chakra-core#1255 from suwc:build/suwc/bugfix

Change default constructor to

constructor(...args) { super(...args); }

from (functionally equivalent pre-spread implementation)

constructor() { super(...arguments); }

Fixes chakra-core#1228
…d includes -ESObjectGetOwnPropertyDescriptors to toggle its availability.
…opertyDescriptors.

Merge pull request chakra-core#1202 from jordonwii:issue254_getownpropertydescriptors

Fixes chakra-core#254

This is a rough implementation based on a combination of getOwnPropertyDescriptor as a reference, and what I could intuit from a day of digging.

I'd appreciate as much feedback as possible on how I did this. I used existing functions I was aware of when possible, but if there are better choices of helper functions here, please do let me know. I also did the tests all in javascript (similar to what I could find on getOwnPropertyDescriptor), and tried to make them as thorough as possible.
…te instead of memset to keep cache untouched

Merge pull request chakra-core#1259 from ThomsonTan:PerfCacheMiss

This is for experimenting non-temporal write when zero-ing pages in background thread, so it doesn't evict cache lines other task is using.
…newline character from function display string

Merge pull request chakra-core#1241 from agarwal-sandeep:removenewline

Chakra  "\nfunction(){\n  [native code]\n}\n"
Firefox "function (){\n}"
Chrome "function () {}"

For arguments.callee.__proto__.toString() (or any native function toString) Chakra shows form feed character (\012) at begining and end whereas Firefox and Chrome doesn't.

VSCode removes the newline after function but doesn't remove the starting so it shows up in locals display.
Merge pull request chakra-core#1182 from rajatd:inlineInJLB-test

This change enables script function inlining in jitted loop bodies.
- Small changes in the stack walker to support this scenario.
- Tweaks to inlining heuristics so that we don't inline as aggressively in loop bodies as we do in functions.
This warning is disabled in other projects, Disabling it in the NativeTests as well.
… and fix warnings in nativetests project

Merge pull request chakra-core#1264 from akroshg:testfix
…rom LeakReport

Merge pull request chakra-core#1265 from digitalinfinity:simplify_leakreport
jenkins script has already determined if it is meant for noJit, so it will pass appropriate location to inner script.
…est script.

Merge pull request chakra-core#1267 from akroshg:testfix1

jenkins script has already determined if it is meant for noJit, so it will pass appropriate location to inner script.

Fixes chakra-core#1266
Cellule added 3 commits July 12, 2016 16:17
…mbly

# Conflicts:
#	bin/ch/ChakraRtInterface.cpp
#	bin/ch/ChakraRtInterface.h
#	bin/ch/Helpers.cpp
#	bin/ch/WScriptJsrt.cpp
#	bin/ch/WScriptJsrt.h
#	lib/Jsrt/ChakraCommon.h
#	lib/Jsrt/Jsrt.cpp
#	lib/Runtime/Base/Chakra.Runtime.Base.vcxproj
#	lib/Runtime/Language/InterpreterLoop.inl
#	lib/Runtime/Library/InJavascript/Intl.js.bc.32b.h
#	lib/Runtime/Library/InJavascript/Intl.js.bc.64b.h
#	lib/Runtime/Library/InJavascript/Intl.js.nojit.bc.32b.h
#	lib/Runtime/Library/InJavascript/Intl.js.nojit.bc.64b.h
#	lib/Runtime/Runtime.h
@Cellule Cellule merged commit ae19857 into chakra-core:WebAssembly Jul 13, 2016
@Cellule Cellule deleted the WebAssembly branch July 25, 2016 20:57
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.