diff options
author | Abseil Team <absl-team@google.com> | 2020-05-05T14·54-0700 |
---|---|---|
committer | vslashg <gfalcon@google.com> | 2020-05-05T17·52-0400 |
commit | d85783fd0b1bb32b3d3e04d18367cec8d96c9e9a (patch) | |
tree | 4ed021579b676d6f387ee071e94451689eb5a868 /absl/random/gaussian_distribution_test.cc | |
parent | a1d6689907864974118e592ef2ac7d716c576aad (diff) |
Export of internal Abseil changes
-- f34cd235a12ad0ee1fea3a1ee5a427272dc2b285 by Abseil Team <absl-team@google.com>: Migrates uses of deprecated map types to recommended types. PiperOrigin-RevId: 309945156 -- e3410a47ad32c0775b6911610bc47b22938decad by Matthew Brown <matthewbr@google.com>: Internal Change PiperOrigin-RevId: 309856021 -- a58cfa25e0bb59e7fa9647ac1aae65eaccff0086 by Greg Falcon <gfalcon@google.com>: Internal change. PiperOrigin-RevId: 309804612 -- cdc5ec310035fbe25f496bda283fe655d94d7769 by Mark Barolak <mbar@google.com>: Standardize the header comments for friend functions in cord.h PiperOrigin-RevId: 309779073 -- fe61602701be795e54477b0fdbf5ffc1df12a6b7 by Samuel Benzaquen <sbenza@google.com>: Implement %f natively for any input. It evaluates the input at runtime and allocates stack space accordingly. This removes a potential fallback into snprintf, improves performance, and removes all memory allocations in this formatting path. PiperOrigin-RevId: 309752501 -- 79e2a24f3f959e8b06ddf1d440bbabbd5f89b5b7 by Greg Falcon <gfalcon@google.com>: Add a Cord::swap() method. Many other Abseil types already provide this, but it was missing here. We already provided a two-argument free function form of `swap()`, but that API is better suited for generic code. The swap member function is a better API when the types are known. PiperOrigin-RevId: 309751740 -- 85cdf60024f153fb4fcb7fe68ed2b14b9faf119d by Derek Mauro <dmauro@google.com>: Cleanup uses of "linker initialized" SpinLocks PiperOrigin-RevId: 309581867 -- 9e5443bfcec4b94056b13c75326576e987ab88fb by Matt Kulukundis <kfm@google.com>: Clarify intended mixing properties of `absl::Hash` PiperOrigin-RevId: 309520174 -- a0630f0827b67f217aaeae68a448fe4c1101e17d by Greg Falcon <gfalcon@google.com>: Comment out a test in Emscripten to sidestep `long double` issues. PiperOrigin-RevId: 309482953 GitOrigin-RevId: f34cd235a12ad0ee1fea3a1ee5a427272dc2b285 Change-Id: Icce0c9d547117374d596b9d684e4054ddd118669
Diffstat (limited to 'absl/random/gaussian_distribution_test.cc')
-rw-r--r-- | absl/random/gaussian_distribution_test.cc | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/absl/random/gaussian_distribution_test.cc b/absl/random/gaussian_distribution_test.cc index 49c07513bf22..398f0131b5dd 100644 --- a/absl/random/gaussian_distribution_test.cc +++ b/absl/random/gaussian_distribution_test.cc @@ -130,12 +130,15 @@ TYPED_TEST(GaussianDistributionInterfaceTest, SerializeTest) { ss >> after; #if defined(__powerpc64__) || defined(__PPC64__) || defined(__powerpc__) || \ - defined(__ppc__) || defined(__PPC__) + defined(__ppc__) || defined(__PPC__) || defined(__EMSCRIPTEN__) if (std::is_same<TypeParam, long double>::value) { // Roundtripping floating point values requires sufficient precision // to reconstruct the exact value. It turns out that long double // has some errors doing this on ppc, particularly for values // near {1.0 +/- epsilon}. + // + // Emscripten is even worse, implementing long double as a 128-bit + // type, but shipping with a strtold() that doesn't support that. if (mean <= std::numeric_limits<double>::max() && mean >= std::numeric_limits<double>::lowest()) { EXPECT_EQ(static_cast<double>(before.mean()), |