fix and enable wasm on xplat #4531
Conversation
MikeHolman
force-pushed
the
fixwasmxplat
branch
2 times, most recently
from
269eb42 to
f1302fa
Compare
MikeHolman
force-pushed
the
fixwasmxplat
branch
2 times, most recently
from
b05cfa6 to
e9b98da
Compare
…uncation. Fix some unit tests to be able to run in xplat builds (mostly problems with filename casing) Use HeapNew & HeapDelete for exception message in Wasm. Make test regress.js more generic to be able to add more regression tests there.
MikeHolman
force-pushed
the
fixwasmxplat
branch
from
e9b98da to
fefa599
Compare
test/wasm/signextend.js
Outdated
| @@ -6,7 +6,7 @@ | |||
| /* global assert,testRunner */ // eslint rule | |||
| WScript.Flag("-WasmSignExtends"); | |||
| WScript.Flag("-WasmI64"); | |||
| WScript.LoadScriptFile("../UnitTestFrameWork/UnitTestFrameWork.js"); | |||
| WScript.LoadScriptFile("..\\UnitTestFramework\\UnitTestFramework.js"); | |||
There was a problem hiding this comment.
The change for the casing is necessary, however, we should still use forward slashes
test/wasm/binary.js
Outdated
| @@ -4,7 +4,7 @@ | |||
| //------------------------------------------------------------------------------------------------------- | |||
|
|
|||
| /* global assert,testRunner */ // eslint rule | |||
| WScript.LoadScriptFile("../UnitTestFrameWork/UnitTestFrameWork.js"); | |||
| WScript.LoadScriptFile("../UnitTestFramework/UnitTestFramework.js"); | |||
| WScript.LoadScriptFile("../wasmspec/testsuite/harness/wasm-constants.js"); | |||
There was a problem hiding this comment.
Instead of wasmspec use WasmSpec or rename the folder.
|
I'm happy that we are about to enable wasm on xplat! @MikeHolman @Cellule Thanks for this! couple of small comments LGTM! |
|
Reviewed 9 of 11 files at r1, 7 of 9 files at r2, 1 of 2 files at r3, 6 of 6 files at r4. Comments from Reviewable |
test/AsmJs/rlexe.xml
Outdated
| @@ -522,6 +524,8 @@ | |||
| <default> | |||
| <files>MathBuiltinsCall.js</files> | |||
| <baseline>MathBuiltinsCall.baseline</baseline> | |||
| <!-- mac gives has some test which gives different precision result, so exclude it --> | |||
There was a problem hiding this comment.
nit: open an issue to track these?
There was a problem hiding this comment.
I can, but do you think it's worth tracking? this is difference in precision with result of atan, which spec says is implementation defined approximation.
There was a problem hiding this comment.
I see. I thought only some tests have difference but others don't. If this is not the case, well, you are right, no point tracking it!
Generally speaking;
What we did with some of the date time and number formatting testing was to enable-disable per platform. i.e. CH supports the line below that checks if the system running the test is a macOS.
var isMac = (WScript.Platform && WScript.Platform.OS == 'darwin');
fix issues with calling convention: we are using stack based calling convention, so we can cleanup a lot of code that was complicated by xmm registers.
fix issue where JIT was calling wrong method for floorf and ceilf
Resolves #3561