Add one-time randomization support for FieldTrial, and the ability to

base/rand_util.cc

Index: base/rand_util.cc
diff --git a/base/rand_util.cc b/base/rand_util.cc
index 4a455f1bd9e2a29d3aae27c6b6d9911c7ea401ec..2a75d31c78a912a8ff4ba4ceeaa3cf69025fcab0 100644
--- a/base/rand_util.cc
+++ b/base/rand_util.cc
@@ -1,4 +1,4 @@
-// Copyright (c) 2010 The Chromium Authors. All rights reserved.
+// Copyright (c) 2011 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
@@ -23,6 +23,10 @@ int RandInt(int min, int max) {
 }
 
 double RandDouble() {
+  return BitsToRandLikeDouble(base::RandUint64());
+}
+
+double BitsToRandLikeDouble(uint64 bits) {

[jar (doing other things), 2011/04/21 22:10:02]

Why not just BitsToDouble()?  I really don't like Rand in there.  Perhaps
BitsToSmallDouble()?  Does that help with your desire to state that there is a
"limited output range," and not merely something like a reintepret_cast?

[Jói, 2011/04/28 01:03:50]

This is technically one of the half-open unit intervals (the other is (0.0,
1.0]).  I am now calling it BitsToOpenEndedUnitInterval which I think captures
fairly nicely that it's the [0.0, 1.0) half-open unit interval.

   // We try to get maximum precision by masking out as many bits as will fit
   // in the target type's mantissa, and raising it to an appropriate power to
   // produce output in the range [0, 1).  For IEEE 754 doubles, the mantissa
@@ -30,7 +34,7 @@ double RandDouble() {
 
   COMPILE_ASSERT(std::numeric_limits<double>::radix == 2, otherwise_use_scalbn);
   static const int kBits = std::numeric_limits<double>::digits;
-  uint64 random_bits = base::RandUint64() & ((GG_UINT64_C(1) << kBits) - 1);
+  uint64 random_bits = bits & ((GG_UINT64_C(1) << kBits) - 1);
   double result = ldexp(static_cast<double>(random_bits), -1 * kBits);
   DCHECK(result >= 0.0 && result < 1.0);
   return result;