You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think there are other mistakes in this comment as well. The largest float32 exponent is 127, thus the largest float32 is maxMant * 2**127. maxMant is 1 (implicit) + (2**23 -1)/(2**23) == 1 + 1 - 1/(2**23). Together we have maxFloat32 == (1 + 1 - 1/(2**23)) * 2**127 == (2 - 2**-23) * 2**127 == 2**128 - 2**104 == f1 (first line).
With f2 == 2**128, f1 + f2 == 2**128 - 2**104 + 2**128 == 2**129 - 2**104. Thus (f1 + f2)/2 == 2**128 - 2**103, as you suggest.
Feel free to send me a CL and assign it to me for review. Thanks.
Please answer these questions before submitting your issue. Thanks!
go version
)?tip @ 0cff219
go env
)?N/A
test/float_lit2:21 reads
// The halfway point is (f₁+f₂)/2 = 2¹²⁸ - 2¹⁰⁵.
The expression should be
2^128-2^103
, ie.s/105/103/
above.If approved, I can send a CL later today.
The text was updated successfully, but these errors were encountered: