RESOLVED FIXED 164611
Move Wasm tests to JS
https://bugs.webkit.org/show_bug.cgi?id=164611
Summary Move Wasm tests to JS
Keith Miller
Reported 2016-11-10 13:16:12 PST
Move Wasm tests to JS
Attachments
Patch (66.72 KB, patch)
2016-11-10 13:16 PST, Keith Miller
no flags
Patch (68.56 KB, patch)
2016-11-11 14:33 PST, Keith Miller
no flags
Patch (95.50 KB, patch)
2016-11-11 14:40 PST, Keith Miller
no flags
Patch (98.31 KB, patch)
2016-11-11 15:53 PST, Keith Miller
no flags
Patch for landing (98.31 KB, patch)
2016-11-11 15:54 PST, Keith Miller
no flags
Keith Miller
Comment 1 2016-11-10 13:16:42 PST
Keith Miller
Comment 2 2016-11-10 13:17:10 PST
Comment on attachment 294402 [details] Patch Shouldn't have been marked for review.
WebKit Commit Bot
Comment 3 2016-11-10 13:19:18 PST
Attachment 294402 [details] did not pass style-queue: ERROR: Source/JavaScriptCore/jsc.cpp:2448: One line control clauses should not use braces. [whitespace/braces] [4] Total errors found: 1 in 16 files If any of these errors are false positives, please file a bug against check-webkit-style.
JF Bastien
Comment 4 2016-11-10 13:47:31 PST
Comment on attachment 294402 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=294402&action=review This is pretty great! > Source/JavaScriptCore/jsc.cpp:1163 > + addFunction(vm, "testWasmModule", functionTestWasmModule, 0); What does it test about a wasm module? I think the name could be improved. > Source/JavaScriptCore/jsc.cpp:2377 > + sscanf(bitwise_cast<const char*>(jsCast<JSString*>(value)->value(exec).characters8()), "%lld", &result); Check return isn't EOF. > Source/JavaScriptCore/jsc.cpp:2387 > + if (typeString == "float") Wouldn't "f32" be more consistent here? > Source/JavaScriptCore/jsc.cpp:2390 > + return value; RELEASE_ASSERT(typeString == "f64"); > Source/JavaScriptCore/testWasm.cpp:464 > + // )I ? > Source/JavaScriptCore/wasm/WasmB3IRGenerator.cpp:60 > +const bool verbose = true; :-) > JSTests/wasm/Builder.js:172 > + } Weird indent. > JSTests/wasm/Builder.js:178 > + assert.eq(imms.length, imm.length, `"${op}" expects ${imm.length} immediates, got ${imms.length}`); What's "imm" here? > JSTests/wasm/Builder.js:235 > + print(nextBuilder) Semicolon. > JSTests/wasm/Builder.js:371 > + Memory: (initial, max) => s.data.push({ initial, max }) I think it would be better to spell out initial and max when used, instead of relying on order here. Maybe have Memory().InitialMaxPages(x, y).End() ? We'll also allow more than 1 in the future, so the API has to support more than one memory. > JSTests/wasm/Builder_WebAssemblyBinary.js:76 > + put(bin, "varuint32", memory.max ? 1 : 0); Comment to explain what this is: // flags > JSTests/wasm/LowLevelBinary.js:231 > + print(this._buffered_bytes); Nope!
Keith Miller
Comment 5 2016-11-11 14:33:38 PST
Keith Miller
Comment 6 2016-11-11 14:40:34 PST
Geoffrey Garen
Comment 7 2016-11-11 14:42:01 PST
Comment on attachment 294541 [details] Patch We want one test per file.
Keith Miller
Comment 8 2016-11-11 15:53:51 PST
Keith Miller
Comment 9 2016-11-11 15:54:54 PST
Created attachment 294556 [details] Patch for landing
WebKit Commit Bot
Comment 10 2016-11-11 16:32:43 PST
Comment on attachment 294556 [details] Patch for landing Clearing flags on attachment: 294556 Committed r208627: <http://trac.webkit.org/changeset/208627>
WebKit Commit Bot
Comment 11 2016-11-11 16:32:47 PST
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.