Export of internal Abseil changes

--
5f3c139695d5c497ca030e95a607537a7be7caa7 by Benjamin Barenblat <bbaren@google.com>:

Don’t examine irrelevant destination buckets in DiscreteDistributionTest

Abseil generates discrete distributions using Walker’s aliasing
algorithm. This creates uniformly distributed buckets, each with a
probability of sending traffic to a different bucket. Abseil represents
a bucket as a pair

    (probability of retaining traffic ×
     alternate bucket if traffic is passed)

and a distribution as a vector of such pairs. For example, {(0.3, 1),
(1.0, 1)} represents a distribution with two buckets, the zeroth of
which passes 70% of its traffic to bucket 1 and the first of which holds
on to all its traffic.

This representation is not unique: When a bucket retains traffic with
probability 1, the alternate bucket is irrelevant. Continuing the
example above, {(0.3, 1), (1.0, 0)} _also_ represents a two-bucket
distribution where the zeroth bucket passes 70% of its traffic to the
first and the first hangs on to all traffic. Exactly what representation
Abseil generates for a given input is related to how much precision is
used in intermediate floating-point operations, which is an
architectural implementation detail. Remove sensitivity to that detail
by not examining the alternate bucket when the retention probability is
1.0.

PiperOrigin-RevId: 372993410

--
062ac80699f748831c09a061538abffec2cdea5c by Martijn Vels <mvels@google.com>:

Avoid alredy sampled cord remaining sampled if not picked or source is sampled

PiperOrigin-RevId: 372985990

--
a9f3537e1110b7bb6450fd72a03f0c5dc6b8c89b by Evan Brown <ezb@google.com>:

Add tests for function pointer comparators, comparators that have SFINAE-visible comparison operators that are unimplemented, and for implicit construction from unadapted comparators.

PiperOrigin-RevId: 372927616
GitOrigin-RevId: 5f3c139695d5c497ca030e95a607537a7be7caa7
Change-Id: I996a8452e7bd88f9dd2e59633b01bbc09f42620d

1 files changed, 17 insertions, 0 deletions

diff --git a/absl/strings/cord.cc b/absl/strings/cord.cc
index 238532f9..5dad781e 100644
--- a/absl/strings/cord.cc
+++ b/absl/strings/cord.cc
@@ -535,6 +535,23 @@ void Cord::InlineRep::AssignSlow(const Cord::InlineRep& src) {
     EmplaceTree(CordRep::Ref(src.as_tree()), src.data_, method);
     return;
   }
+
+  // See b/187581164: unsample cord if already sampled
+  // TODO(b/117940323): continuously 'assigned to' cords would reach 100%
+  // sampling probability. Imagine a cord x in some cache:
+  //   cache.SetCord(const Cord& foo) {
+  //     x = foo;
+  //   }
+  // CordzInfo::MaybeTrackCord does:
+  // x.profiled = foo.profiled | x.profiled | random(cordz_mean_interval)
+  // Which means it will on the long run converge to 'always samples'
+  // The real fix is in CordzMaybeTrackCord, but the below is a low risk
+  // forward fix for b/187581164 and similar BT benchmark regressions.
+  if (ABSL_PREDICT_FALSE(is_profiled())) {
+    cordz_info()->Untrack();
+    clear_cordz_info();
+  }
+
   CordRep* tree = as_tree();
   if (CordRep* src_tree = src.tree()) {
     data_.set_tree(CordRep::Ref(src_tree));