math/rand: problem with seed on windows/amd64

[gopherbot]

by kpashka:

> What does 'go version' print?

go version go1.3.3 windows/amd64

> What steps reproduce the problem?

http://play.golang.org/p/-JoqAkDZ16

> What happened?

First three numbers are same, fourth is different from three previous.

> What should have happened instead?

All numbers should (not always, but often) be different.

> Please provide any additional information below.

I know that Go Playground has special environment, but while running this test on
Windows 8.1 x64, I get the same incorrect results. My co-worker have ran this on Gentoo
(go version 1.3 linux/amd64) and got different numbers. I've also tried this using
windows/386 version and got incorrect results.

[cznic]

Comment 1:

Windows have poor time granularity, thus the first three calls may seed the PRNG with
the same number. You need to use some other source of entropy or seed the PRNG only
once, if possible.

[bradfitz]

Comment 2:

Status changed to WorkingAsIntended.

[gopherbot]

Comment 3 by kpashka:

Ruby (and many other languages) use RSA cryptographic provider on Windows while seeding
RNG, e.g. http://www.ruby-doc.org/core-2.1.3/Random.html
Why this isn't possible in golang?