code review 4342043: crypto/rand: Fix on Plan 9

crypto/rand: Fix on Plan 9 Takes the currently unused prng code from rand_unix.go and puts it into a prng.go. Created a rand_plan9.go which uses /dev/random and prng.go. The dev_readers are not duplicate since the Plan 9 one lacks the bufio (it would be bad to try to read a whole buffer from /dev/random).

[taruti, 2011-04-02 16:22:34]

Please review.

[fhs, 2011-04-03 05:07:45]

http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/prng.go File src/pkg/crypto/rand/prng.go (right): http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/prng.go#n... src/pkg/crypto/rand/prng.go:17: // haven't needed it. This second sentence should go away. http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/rand_plan... File src/pkg/crypto/rand/rand_plan9.go (right): http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/rand_plan... src/pkg/crypto/rand/rand_plan9.go:14: Everything after this line is duplicated in rand_unix.go. You can move it to rand.go. http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/rand_unix.go File src/pkg/crypto/rand/rand_unix.go (left): http://codereview.appspot.com/4342043/diff/2001/src/pkg/crypto/rand/rand_unix... src/pkg/crypto/rand/rand_unix.go:16: "time" time doesn't compile yet (at least for me). Better wait until that's done.

[taruti, 2011-04-03 16:33:06]

Hello golang-dev@googlegroups.com, fhs (cc: golang-dev@googlegroups.com, paulzhol), I'd like you to review this change to https://go.googlecode.com/hg/

[paulzhol, 2011-04-03 16:44:42]

Fhs has a point about the duplicated parts and pkg time is not ready yet

[taruti, 2011-04-03 18:53:46]

On Sun, 03 Apr 2011 16:44:42 +0000, paulzhol@gmail.com wrote: > Fhs has a point about the duplicated parts and pkg time is not ready yet If you look at the code they are not duplicate (although that could be made clearer). Reading from /dev/urandom (unix) uses a buffered reader, while reading from /dev/random should not use a buffered reader since that would incur long waits. But I agree that before committing time should be ready. - Taru Karttunen 

[paulzhol, 2011-04-03 19:18:34]

On 2011/04/03 18:53:46, taruti wrote: > If you look at the code they are not duplicate (although that could be > made clearer). Reading from /dev/urandom (unix) uses a buffered reader, > while reading from /dev/random should not use a buffered reader since > that would incur long waits. > > But I agree that before committing time should be ready. Sorry, I missed the r.f = bufio.NewReader(f) line. Nonetheless the devReader type declaration is duplicated. p.s what is the rational for buffering reads from /dev/urandom ?

[ality, 2011-04-04 09:22:46]

LGTM, once my time changeset is submitted.

[taruti, 2011-04-06 05:44:01]

Hello golang-dev@googlegroups.com, fhs, paulzhol, ality (cc: golang-dev@googlegroups.com), Please take another look.

[ality, 2011-06-20 08:10:08]

On 2011/04/06 05:44:01, taruti wrote: > Please take another look. Hey Taru, can you sync this to tip? Thanks, Anthony

[taruti, 2011-06-21 07:23:50]

Hello golang-dev@googlegroups.com, fshahriar@gmail.com, paulzhol@gmail.com, ality@pbrane.org (cc: golang-dev@googlegroups.com), Please take another look.

[ality, 2013-07-20 06:49:04]

This appears to have been fixed slightly differently on June 6th, 2012 (changeset c94261e6c85c). Moving reviewers to CC.