We have code for converting C/C++/Objc keywords that appear in protos
to convert them so that they can be compiled.
One of the things we need to be careful of is accidentally overriding methods
that Apple declares in NSObject. It turns out that we have run into issues
where we conflict with "hidden" methods in NSObject or methods added by
categories. method_dump.sh collects all of the methods we care about for
macOS and iOS and dumps them into objectivec_nsobject_methods.h which
is then included in objectivec_helpers.cc as part of the build.
Added a pile of tests to verify that conversions are happening as expected.
This unbreaks the Windows shared-library build of Chrome when using
Clang versions based on r344987 or later, see https://crbug.com/901776
implicit_weak_message.cc is part of protobuf_lite.dll, and it includes
wire_format_lite.h, which includes the dllexport inline function
WireFormatLite::WriteGroupToArray which will therefore be emitted.
WriteGroupToArray in turn calls the inline function
InternalWriteGroupToArray, however that definition is provided in the
_inl file. To make sure the definition is available, the _inl file must
be included.
Before Clang r344987 the build worked anyway due to luck, because
InternalWriteGroupToArray was emitted into other object files (e.g. in
wire_format_lite.obj). After that Clang revision, those definitions
started getting inlined, and so are longer available and cause a link
failure for the reference from implicit_weak_message.obj.
No changes were needed, but since the Xcode projects pick up the updated
setting, the tests require a newer Xcode that supports Swift 4.
This is being done because Xcode 10 starting warning about Swift 3 support
going away in the future, so we might as well do the updates since most
folks shouldn't be on those really old Xcode versions any more.
Not sure the history for why there were two objc_library targets, but
given the one has an issue about not working, merge them into a single
target and an alias.
- Use the alias for :objectivec since that naming doesn't seem to follow
the other targets *_library targets here.
- Update :protobuf_objc to be use for the working target and just inline
the sources/headers lists so there is less indirection in the file.
Fixes#5284
* Replace assert_true/assert_false with assert, because they are not provided in newer version of test/unit
* Use rescue to handle exception subclass
* Commented out the require that caused gc_test failure.
* Commented out test
* Fix typo
* Add support for primitive types in setters
* Update to address PR feedback
* Add tests and fixes for repeated fields
* Remove repeated field code, add getters
* Cleanup, test getters and oneofs
* Move boxing logic into separate class
* Add tests for wrapper type constructor args
* Update to add new setXXXValue methods
* Fix tests for invalid values
* Fix c extension for wrapper accessors
* Fix the bug that well known types didn't call Message_construct
* Address PR comments
* Refactoring init message with array logic
* Add include path to protoc
* Add missing TSRM_LS defintion
* Fix TSRM_LS
* Fix dist check
* Fix parsing empty Struct Values from Json
This fixes a bug. When parsing a struct from JSON like
struct = json_format.Parse('{"k": {}}', Struct())
then the struct's "k" value would end up not initialized, and accessing
the value would raise an error.
In[1]: struct['k']
ValueError: Value not set
That seems to be because the Struct field of the Value was not set.
In[2]: struct
Out[2]:
fields {
key: "k"
value {
}
}
This commit makes sure that the Value's Struct field is initialized even
if the Struct has no values itself.
This commit also extends a test to cover this case.
* Additionally test for empty list
* ParseFromString returns bytes parsed
ParseFromString is documented as returning the number of bytes parsed,
and the C++ implementation does this, so the Python implementation
should too.
See #5165.
* Fix docstring
* Add a test for ParseFromString length behavior