Updates:
Status: Fixed
Owner: yang...@chromium.org
Comment #9 on issue 3035 by yang...@chromium.org: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
Fixed in r22918
--
You received this message because this project is configured to send all
is
Comment #8 on issue 3035 by r...@chromium.org: Math.sin(Math.PI) inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
I think what Kahan says in http://www.cs.berkeley.edu/~wkahan/JAVAhurt.pdf
is relevant. See page 25, section 13:
13 Prevalent Misconceptions about Floating-Point Arit
Comment #7 on issue 3035 by svenpa...@chromium.org: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
Every argument made about floating point and accuracy which doesn't involve
the notion of ranges if basically invalid. Every floating point value does
not rep
Comment #6 on issue 3035 by r...@chromium.org: Math.sin(Math.PI) inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
No, the expected answer cannot not be zero because Math.PI is NOT pi.
--
You received this message because this project is configured to send all
issue notifications t
Comment #5 on issue 3035 by oao.hika...@gmail.com: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
0.0001224646798818428 (actual)
0.000 (expected)
(1.224646798818428e-16 - 0) <= 1e-15 // true
--
You received this messag
Comment #4 on issue 3035 by r...@chromium.org: Math.sin(Math.PI) inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
I filed a bug because the current value is only accurate to about 10
digits, out of about 15. The C library produces 15 digits.
--
You received this message becaus
Comment #3 on issue 3035 by jkumme...@chromium.org: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
OK, maybe I wasn't clear.
My first statement commented on the fact that the new result is much closer
to what you stated to be the correct answer. Given this,
Comment #2 on issue 3035 by r...@chromium.org: Math.sin(Math.PI) inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
Your two statements are inconsistent. You can't claim that it's better and
also believe abs(expected-actual) <= 1e-15 is the best you can hope for.
And I don't under
Updates:
Status: WorkingAsIntended
Comment #1 on issue 3035 by jkumme...@chromium.org: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
So precision got much better with the new implementation, that's great!
FWIW, in my opinion Math.sin(Math.PI) should b
