From es-discuss: https://mail.mozilla.org/pipermail/es-discuss/2014-September/039470.html Related v8 issue: https://code.google.com/p/v8/issues/detail?id=3579 Expected: No output Actual: 441 powers of two have inexact results, e.g. `Math.log2(Math.pow(2, 124))` outputs `124.00000000000001` instead of `124` Reproducible on Win7 32bit, Fx32. Depends on whether or not log2 is supported by the compiler, see jsmath.cpp. for (var k = -1074; k <= 1023; ++k) { if (Math.log2(Math.pow(2, k)) != k) { console.log("Math.log2(2^" + k + "): "+ Math.log2(Math.pow(2, k))) } }

// extracts the exponent from the double value Math.getExponent = function (x) { var e = Math.floor(Math.log(x) / Math.log(2) + 0.5); return Math.pow(2, e) > x ? e - 1 : e; }; Math.log2 = function (x) { if (x === 0) { return -1 / 0; } if (x === 1 / 0) { return 1 / 0; } var e = Math.getExponent(x); if (e < 0) { e += 1; } return Math.log(x / Math.pow(2, e)) / Math.log(2) + e; };

Now the testcase in comment #0 don't show anything, on Winxp 32bit nightly (2016-07-04), as bug 933257 changed the implementation to fdlibm. we could add the testcase and close this bug :)

Just increased the range. it passes on all archs https://treeherder.mozilla.org/#/jobs?repo=try&revision=b247807c029b

Comment on attachment 8769297 [details] [diff] [review] Increase the testing range for Math.log2 + Math.pow. Review of attachment 8769297 [details] [diff] [review]: ----------------------------------------------------------------- Bleh, sorry for the long delay - I somehow didn't see this :(

https://hg.mozilla.org/integration/mozilla-inbound/rev/016d82db1c6b20bdefeb8093e6320bc10b8c0f37 Bug 1070517 - Increase the testing range for Math.log2 + Math.pow. r=till