Math tests from https://github.com/mozilla/mozilla-central #85
Conversation
|
I suppose all of them are failing. Right? |
|
no, 2435 passes, 166 failures in Chrome and Firefox |
|
after applyling fixes for Math.trunc and Math.hypot - #35 (comment) - 153 failures |
|
okay then, let’s leave this opened and incrementally improve shim to make all tests pass |
| assertEq(Math.sign(-Number.MAX_VALUE), -1); | ||
| assertEq(Math.sign(-Infinity), -1); | ||
|
|
||
| for (var i = -1 |
There was a problem hiding this comment.
please add a trailing newline - omitting one screws with many unix tools, including git diff.
|
How many of these tests are in the spec? We should only include tests that are in the ES6 spec - not just tests that a given ES implementation uses - as the behavior will differ between V8, SpiderMonkey, Trident, Blink, etc. |
|
@ljharb, but do we trying to reduce this difference or not? |
|
@Yaffle the spec is not very specific on many things - please go take a look at it. There is a danger from over-specifying standard methods, and we shouldn't prefer one engine over any other, or else people won't want to use es6-shim across multiple engines. If the spec allows for deviation, and V8 deviates much more than mozilla, they're both correct - so a test would be meaningless. |
|
@ljharb, please take a look at tests - there is only math, nothing engine-specific, excepp Number.MIN_VALUE |
|
@Yaffle if the values aren't specified in the spec, they're potentially engine-specific. Sanity checking is awesome - we just need to be careful that tests aren't too precise, because if it's not spec'd, there is no right answer. |
No description provided.