133924 2014-06-14 23:05:59 -0700 Bracket access is significantly slower than property access 2014-06-14 23:23:46 -0700 1 1 1 Unclassified WebKit JavaScriptCore 528+ (Nightly build) Unspecified Unspecified NEW P2 Enhancement --- 1 rniwa webkit-unassigned barraclough fpizlo kling mhahnenberg mike oldest_to_newest 1015751 0 233135 rniwa 2014-06-14 23:05:59 -0700 Created attachment 233135 Microbenchmark function getByVal(object, property) { return object[property]; } function getById(object) { return object.myProperty; } The first function appears to be ~3.5x slower than the second function even if getByVal is always called as getByVal(object, 'myProperty'). It's only 1.5x slower on Chrome 35 and 2.2x slower on Firefox 30. 1015752 1 rniwa 2014-06-14 23:07:14 -0700 Note that it's 3.7x slower in the shipping Safari so it has gotten better. 1015753 2 rniwa 2014-06-14 23:07:27 -0700 (In reply to comment #1) > Note that it's 3.7x slower in the shipping Safari so it has gotten better. 4.7x* 233135 2014-06-14 23:05:59 -0700 2014-06-14 23:05:59 -0700 Microbenchmark getbyval-getbyid.html text/html 1052 rniwa PCFET0NUWVBFIGh0bWw+CjxodG1sPgo8Ym9keT4KPHNjcmlwdD4KCndpbmRvdy5yZXN1bHQgPSBu dWxsOwoKZnVuY3Rpb24gZ2V0QnlWYWwob2JqZWN0LCBwcm9wZXJ0eSkgewogICAgcmV0dXJuIG9i amVjdFtwcm9wZXJ0eV07Cn0KCmZ1bmN0aW9uIGdldEJ5SWQob2JqZWN0KSB7CiAgICByZXR1cm4g b2JqZWN0Lm15UHJvcGVydHk7Cn0KCmZ1bmN0aW9uIHJ1blRlc3QodGVzdEZ1bmN0aW9uKSB7CiAg ICB2YXIgcmVzdWx0ID0gMDsKICAgIHZhciBzdGFydCA9IERhdGUubm93KCk7CiAgICB2YXIgb2Jq ZWN0ID0ge215UHJvcGVydHk6IDEyM307CiAgICBmb3IgKHZhciBqID0gMDsgaiA8IDEwMDAgKiAx MDAwOyBqKyspCiAgICAgICAgcmVzdWx0ICs9IHRlc3RGdW5jdGlvbihvYmplY3QsICdteVByb3Bl cnR5Jyk7CiAgICByZXR1cm4gRGF0ZS5ub3coKSAtIHN0YXJ0Owp9Cgp2YXIgcmVzdWx0cyA9IFtd OwoKZnVuY3Rpb24gY29tcGFyZShjb3VudCkgewogICAgdmFyIGRvY1Jlc3VsdCA9IHJ1blRlc3Qo Z2V0QnlWYWwpOwogICAgdmFyIGRvY0VsZW1lbnRSZXN1bHQgPSBydW5UZXN0KGdldEJ5SWQpOwog ICAgcmVzdWx0cy5wdXNoKGRvY1Jlc3VsdCAvIGRvY0VsZW1lbnRSZXN1bHQpOwogICAgZG9jdW1l bnQuYm9keS5pbm5lckhUTUwgKz0gcmVzdWx0c1tyZXN1bHRzLmxlbmd0aCAtIDFdLnRvUHJlY2lz aW9uKDQpICsgJzxicj4nOwoKICAgIGNvdW50LS07CiAgICBpZiAoY291bnQpCiAgICAgICAgc2V0 VGltZW91dChjb21wYXJlLmJpbmQodGhpcywgY291bnQpLCAxMDApOwogICAgZWxzZSB7CiAgICAg ICAgZG9jdW1lbnQuYm9keS5pbm5lckhUTUwgKz0gJ0F2ZXJhZ2U6ICcKICAgICAgICAgICAgKyAo cmVzdWx0cy5yZWR1Y2UoZnVuY3Rpb24gKGEsIGIpIHsgcmV0dXJuIGEgKyBiOyB9KSAvIHJlc3Vs dHMubGVuZ3RoKS50b1ByZWNpc2lvbig0KTsKICAgIH0KfQoKc2V0VGltZW91dChjb21wYXJlLmJp bmQodGhpcywgMjApLCAxMDApOwoKPC9zY3JpcHQ+Cjxmb3JtIG5hbWU9ImdldEVsZW1lbnRCeUlk Ij48L2Zvcm0+CjwvYm9keT4KPC9odG1sPgo=