MINOR: [C++] Avoid multimap::find unspecified behavior

[rupprecht]

Rationale for this change

When a std::multimap has multiple entries with the same key, calling m.find(key) returns an unspecified element.

Historically, this returns the first matching element. However, this is not guaranteed, and recent libc++ changes make this return an arbitrary element. This can lead to surprising behavior, i.e. different values returned based on what other entries are in the multimap (and what the shape of the internal tree is).

What changes are included in this PR?

Replace m.find(key) with m.equal_range(key) to preserve behavior and have predictable results. The behavior of this is guaranteed to return a range of all matching elements in insertion order, and the beginning of the range is the same element as what's normally returned by m.find(key).

Are these changes tested?

Yes

Are there any user-facing changes?

No

[lidavidm]

Thanks!

[lidavidm]

It appears this needs to be formatted:

--- a/cpp/src/arrow/flight/transport/grpc/util_internal.cc
+++ b/cpp/src/arrow/flight/transport/grpc/util_internal.cc
@@ -55,30 +55,25 @@ static bool FromGrpcContext(const ::grpc::ClientContext& ctx,
   const std::multimap<::grpc::string_ref, ::grpc::string_ref>& trailers =
       ctx.GetServerTrailingMetadata();
 
-  const auto [code_val_begin, code_val_end] =
-      trailers.equal_range(kGrpcStatusCodeHeader);
+  const auto [code_val_begin, code_val_end] = trailers.equal_range(kGrpcStatusCodeHeader);
   if (code_val_begin == code_val_end) return false;
 
   std::optional<std::string> message;
-  if (const auto [it, end] = trailers.equal_range(kGrpcStatusMessageHeader);
-      it != end) {
+  if (const auto [it, end] = trailers.equal_range(kGrpcStatusMessageHeader); it != end) {
     message = std::string(it->second.data(), it->second.size());
   }
 
   std::optional<std::string> detail_message;
-  if (const auto [it, end] = trailers.equal_range(kGrpcStatusDetailHeader);
-      it != end) {
+  if (const auto [it, end] = trailers.equal_range(kGrpcStatusDetailHeader); it != end) {
     detail_message = std::string(it->second.data(), it->second.size());
   }
 
   std::optional<std::string> detail_bin;
-  if (const auto [it, end] = trailers.equal_range(kBinaryErrorDetailsKey);
-      it != end) {
+  if (const auto [it, end] = trailers.equal_range(kBinaryErrorDetailsKey); it != end) {
     detail_bin = std::string(it->second.data(), it->second.size());
   }
 
-  std::string code_str(code_val_begin->second.data(),
-                       code_val_begin->second.size());
+  std::string code_str(code_val_begin->second.data(), code_val_begin->second.size());
   *status = internal::ReconstructStatus(code_str, current_status, std::move(message),
                                         std::move(detail_message), std::move(detail_bin),
                                         std::move(flight_status_detail));

[rupprecht]

It appears this needs to be formatted:

Oops, fixed.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 8040f2a.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details.