All TypedArray jsapi-test tests currently work with ArrayBuffer. Clearly we need to do better once TypedArray can map SharedArrayBuffer.

The "p" failure in that run is bug 1230162, technically a problem with running tests in parallel but fundamentally we should also clean up SpiderMonkey.

Comment on attachment 8695789 [details] [diff] [review] generalize jsapi-tests for SAB Review of attachment 8695789 [details] [diff] [review]: ----------------------------------------------------------------- ::: js/src/jsapi-tests/testTypedArrays.cpp @@ +66,5 @@ > > return ok; > } > > +// Shared memory can only be mapped by a TypedArray by creating the TypedArray with "with" should be moved to the next line, to make this line fit into 79 chars. @@ +115,5 @@ > TestArrayFromBuffer(JSContext* cx) > { > size_t elts = 8; > size_t nbytes = elts * sizeof(Element); > + RootedObject buffer(cx, Shared ? JS_NewSharedArrayBuffer(cx, nbytes) : JS_NewArrayBuffer(cx, nbytes)); please wrap this line into 99 chars ;) (or maybe there's different style rule for test files?) @@ +120,5 @@ > { > JS::AutoCheckCannotGC nogc; > bool isShared; > + void* data = Shared ? JS_GetSharedArrayBufferData(buffer, &isShared, nogc) > + : JS_GetArrayBufferData(buffer, &isShared, nogc); ":" should be aligned to the same column as "S" of "Shared ?" in previous line. @@ +121,5 @@ > JS::AutoCheckCannotGC nogc; > bool isShared; > + void* data = Shared ? JS_GetSharedArrayBufferData(buffer, &isShared, nogc) > + : JS_GetArrayBufferData(buffer, &isShared, nogc); > + MOZ_ASSERT(Shared == isShared); as "Shared" is now variable, using CHECK might be better for testing purpose here, and following several places.

https://hg.mozilla.org/integration/mozilla-inbound/rev/a42590c2a8bf36bc766621cb0362dff28de037df Bug 1225040 - generalize jsapi-tests for SAB. r=arai