If messages A and B have the same oneof case, which is a message
type, and we merge B into A, those sub-messages should be merged.
Fixes#3200.
Note that I haven't regenerated all the code, as some of the protos
have been changed, breaking generation.
* Reserve unknown in ruby
* Revert ruby tests. Wait for cpp impl for conformance test
* Add conformance test for preserving unknown
* Add unknown field conformance test to csharp failure list.
* Fix comments
* Fix comment
* Fix comments
* Fix typo
* Use stringsink_string directly
* Mark hd unused
* Remove unused encodeunknown_handlerfunc
* Use constexpr more with VC++ 2017
Chrome's official builds have over 170 dynamic initializers for
variables of the form *::TableStruct::aux. Defining
PROTOBUF_CONSTEXPR_VAR to be constexpr for VS 2017 gets rid of all of
these and saves about 10 KB of binary size.
* Update generated_message_table_driven.h
Restore accidentally deleted line.
Due to https://bugs.llvm.org/show_bug.cgi?id=34198, clang's static
analyzer emits diagnostics about leaking `container`. Doing this
assignment in two steps works around this, and shouldn't cause these
issues.
The typedefs for Atomic32 and Atomic64 were sometimes causing
Atomic32 to be an int32 rather than an intptr_t on 32-bit platforms.
On some of these platforms (ARM/CortexM building with GCC 6 in one
case) int32 is a long int, while intptr_t is an int, which causes a
compiler error even though long int and int are both 4 bytes.
Having Atomic32 always be intptr_t on 32-bit platforms and Atomic64
always be intptr_t on 64-bit platforms should resolve any of these
types of errors.
Jisi Liu
Jon Skeet
xuwei-k
Paul Yang
John Millikin
Tobias Jungel
Francois Marier
Pavel
Bruce Dawson
drivehappy
Adam Cozzette
Bo Yang
Feng Xiao
George Burgess IV
Jisi Liu
Brad Larson