Update webkit-webgl-test-harness.js for more details on WebGL 2 conformance tests
Created attachment 347717 [details] Patch
Comment on attachment 347717 [details] Patch Clearing flags on attachment: 347717 Committed r235206: <https://trac.webkit.org/changeset/235206>
All reviewed patches have been landed. Closing bug.
<rdar://problem/43623888>
Comment on attachment 347717 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=347717&action=review > LayoutTests/webgl/2.0.0/resources/webkit-webgl-test-harness.js:53 > + list(msg, "green"); I think I misled you by suggesting using the <ol> since it doesn't carry into the expected results. I think the output could be improved more. The |msg| reported by the tests are not in the usual format that we're used to reading, so when I read some of these comments, it's hard to tell whether it's a good thing that "default precision qualifier shouldn't work with vec2". And if everything passes, it's hard to tell, despite the final "PASS". Can we reformat these so that they look like resultNum + ": PASS. " + msg Then something like attrib-location-length-limits is easier to read: 1: PASS. gl.getProgramParameter(program, gl.LINK_STATUS) is true 2: FAIL. getError expected: NO_ERROR. Was INVALID_VALUE : 3. FAIL. attrib location was -1, should not be 4: FAIL. getError expected: NO_ERROR. Was INVALID_VALUE : 5: PASS. wtu.loadProgramFromScriptExpectError(gl, "badVertexShader", "fragmentShader") is null 6: PASS. getError was expected value: NO_ERROR : 7: PASS. getError was expected value: INVALID_VALUE : 8: PASS. gl.getAttribLocation(program, attrib1025Name) is -1 9: PASS. getError was expected value: INVALID_VALUE : successfullyParsed is true FAIL Also looking at the tests with expected results that changed, it looks like the "successfullyParsed is true" counts as a PASS? Is that necessary? Can we prevent that?