Export of internal Abseil changes

--
990253454819ce26ff1dda9ab4bbc145b61d01e4 by Xiaoyi Zhang <zhangxy@google.com>:

Import github PR https://github.com/abseil/abseil-cpp/pull/645

PiperOrigin-RevId: 303119797

--
5ac845cb7929b7d1eaf59a309afd811db5001175 by Abseil Team <absl-team@google.com>:

Fix internal exception spec compatibility error

PiperOrigin-RevId: 303104081

--
3290595dd866eecab3c7044e2e3ca0adb74f1bf5 by Gennadiy Rozental <rogeeff@google.com>:

Use FlagValue<T> to represent the value of a flag. Place it directly after
FlagImpl and use a computed offset refer to it.

The offset is computed based on the assumption that the `value_` data member
is placed directly after the impl_ data member in Flag<T>.

This change will allow us to migrate to `T`-specific storage in the generic case.

This change decreases the overhead for int flags by 32 bytes.

PiperOrigin-RevId: 303038099

--
f2b37722cd7a6d3a60ef9713f0d2bbff56f3ddbf by Derek Mauro <dmauro@google.com>:

Minor correctness fix for an ABSL_HAVE_BUILTIN conditional

PiperOrigin-RevId: 302980666

--
39c079a6141ae1c5728af8bf33a39c8aff9deb9f by Abseil Team <absl-team@google.com>:

Use ABSL_HARDENING_ASSERT in b-tree and SwissTable iterators.

PiperOrigin-RevId: 302970075

--
9668a044e080c789df32bcaa1ffb5100831cd9fa by Benjamin Barenblat <bbaren@google.com>:

Correct `add_subdirectory` line in CMake googletest support

Commit bcefbdcdf6ad85046ccacee0aeffba5404d3e528 added support for building with CMake against a local googletest checkout, but I missed a line when constructing the diff. Change the `add_subdirectory` line to reference the correct directories.

PiperOrigin-RevId: 302947488

--
0a3c10fabf80a43ca69ab8b1570030e55f2be741 by Andy Soffer <asoffer@google.com>:

Remove unused distribution format traits.

PiperOrigin-RevId: 302896176

--
0478f2f6270e5ed64c0e28ec09556ca90b2d46a9 by Samuel Benzaquen <sbenza@google.com>:

Fix for CWG:2310.

PiperOrigin-RevId: 302734089

--
3cb978dda5cae5905affdc0914dcc2d27671ed11 by Samuel Benzaquen <sbenza@google.com>:

Fix the Allocate/Deallocate functions to use the same underlying allocator type.

PiperOrigin-RevId: 302721804

--
ae38d3984fb68b4e3ddc165fa8d5c24d5936be52 by Matthew Brown <matthewbr@google.com>:

Internal Change

PiperOrigin-RevId: 302717314

--
7357cf7abd03cc60b6e82b5f28a8e34935c3b4dc by Andy Getzendanner <durandal@google.com>:

Fix typo: s/ABSL_HARDENED_ASSERT/ABSL_HARDENING_ASSERT/

PiperOrigin-RevId: 302532164
GitOrigin-RevId: 990253454819ce26ff1dda9ab4bbc145b61d01e4
Change-Id: Ie595a221c16e1e7e1255ad42e029b646c5f3e11d

1 files changed, 23 insertions, 18 deletions

diff --git a/absl/strings/internal/str_format/float_conversion.cc b/absl/strings/internal/str_format/float_conversion.cc
index d4c647c3ed99..d5a1ee40bc67 100644
--- a/absl/strings/internal/str_format/float_conversion.cc
+++ b/absl/strings/internal/str_format/float_conversion.cc
@@ -33,7 +33,7 @@ bool FallbackToSnprintf(const Float v, const ConversionSpec &conv,
     if (std::is_same<long double, Float>()) {
       *fp++ = 'L';
     }
-    *fp++ = FormatConversionCharToChar(conv.conv());
+    *fp++ = FormatConversionCharToChar(conv.conversion_char());
     *fp = 0;
     assert(fp < fmt + sizeof(fmt));
   }
@@ -100,17 +100,19 @@ bool ConvertNonNumericFloats(char sign_char, Float v,
   char text[4], *ptr = text;
   if (sign_char) *ptr++ = sign_char;
   if (std::isnan(v)) {
-    ptr = std::copy_n(FormatConversionCharIsUpper(conv.conv()) ? "NAN" : "nan",
-                      3, ptr);
+    ptr = std::copy_n(
+        FormatConversionCharIsUpper(conv.conversion_char()) ? "NAN" : "nan", 3,
+        ptr);
   } else if (std::isinf(v)) {
-    ptr = std::copy_n(FormatConversionCharIsUpper(conv.conv()) ? "INF" : "inf",
-                      3, ptr);
+    ptr = std::copy_n(
+        FormatConversionCharIsUpper(conv.conversion_char()) ? "INF" : "inf", 3,
+        ptr);
   } else {
     return false;
   }
 
   return sink->PutPaddedString(string_view(text, ptr - text), conv.width(), -1,
-                               conv.flags().left);
+                               conv.has_left_flag());
 }
 
 // Round up the last digit of the value.
@@ -358,9 +360,9 @@ void WriteBufferToSink(char sign_char, string_view str,
                                        static_cast<int>(sign_char != 0),
                                    0)
                         : 0;
-  if (conv.flags().left) {
+  if (conv.has_left_flag()) {
     right_spaces = missing_chars;
-  } else if (conv.flags().zero) {
+  } else if (conv.has_zero_flag()) {
     zeros = missing_chars;
   } else {
     left_spaces = missing_chars;
@@ -382,9 +384,9 @@ bool FloatToSink(const Float v, const ConversionSpec &conv,
   if (std::signbit(abs_v)) {
     sign_char = '-';
     abs_v = -abs_v;
-  } else if (conv.flags().show_pos) {
+  } else if (conv.has_show_pos_flag()) {
     sign_char = '+';
-  } else if (conv.flags().sign_col) {
+  } else if (conv.has_sign_col_flag()) {
     sign_char = ' ';
   }
 
@@ -401,14 +403,14 @@ bool FloatToSink(const Float v, const ConversionSpec &conv,
 
   Buffer buffer;
 
-  switch (conv.conv()) {
+  switch (conv.conversion_char()) {
     case ConversionChar::f:
     case ConversionChar::F:
       if (!FloatToBuffer<FormatStyle::Fixed>(decomposed, precision, &buffer,
                                              nullptr)) {
         return FallbackToSnprintf(v, conv, sink);
       }
-      if (!conv.flags().alt && buffer.back() == '.') buffer.pop_back();
+      if (!conv.has_alt_flag() && buffer.back() == '.') buffer.pop_back();
       break;
 
     case ConversionChar::e:
@@ -417,9 +419,10 @@ bool FloatToSink(const Float v, const ConversionSpec &conv,
                                                  &exp)) {
         return FallbackToSnprintf(v, conv, sink);
       }
-      if (!conv.flags().alt && buffer.back() == '.') buffer.pop_back();
-      PrintExponent(exp, FormatConversionCharIsUpper(conv.conv()) ? 'E' : 'e',
-                    &buffer);
+      if (!conv.has_alt_flag() && buffer.back() == '.') buffer.pop_back();
+      PrintExponent(
+          exp, FormatConversionCharIsUpper(conv.conversion_char()) ? 'E' : 'e',
+          &buffer);
       break;
 
     case ConversionChar::g:
@@ -446,13 +449,15 @@ bool FloatToSink(const Float v, const ConversionSpec &conv,
         }
         exp = 0;
       }
-      if (!conv.flags().alt) {
+      if (!conv.has_alt_flag()) {
         while (buffer.back() == '0') buffer.pop_back();
         if (buffer.back() == '.') buffer.pop_back();
       }
       if (exp) {
-        PrintExponent(exp, FormatConversionCharIsUpper(conv.conv()) ? 'E' : 'e',
-                      &buffer);
+        PrintExponent(
+            exp,
+            FormatConversionCharIsUpper(conv.conversion_char()) ? 'E' : 'e',
+            &buffer);
       }
       break;