Default to collecting unknown fields as data and then create the
`GPBUnknownFieldSet` on demand when requested.
This should reduce the as the extra objects are created until requested
and clears the way to eventually deprecated `GPBUnknownFieldSet` in the
future.
This also fixes the failing conformance test related to keeping the ordering.
PiperOrigin-RevId: 655929316
We have received several reports in #17036 that the addition of this flag
actually broke the use of command argument files with non-ASCII characters in
their names. It looks like #14253 ended up fixing the original issue with a
different solution anyway. Hopefully this change fixes the issue with non-ASCII
characters.
PiperOrigin-RevId: 655660885
- Add overloads that take `absl::Cord` and `std::string&&` as inputs, putting the burden in the implementation instead of users.
- Add overload that returns `absl::Span<char>` for callers that need higher performance requirements where we can avoid copies altogether.
- Hide the APIs that return `std::string*` when the breaking change is enabled (via `-D PROTOBUF_TEMPORARY_ENABLE_STRING_VIEW_RETURN_TYPE`).
PiperOrigin-RevId: 655600399
- We introduce two new view types ProtoStringCow and ProtoBytesCow.
- In UPB, for cord field accessors we always return a Cow::Borrowed.
- In C++, for coed field accessors we check if the underlying absl::Cord is flat (contigous) and if so return a Cow::Borrowed. If it's not flat we copy the data to a ProtoString and return a Cow::Owned.
- We expect the absl::Cord to be flat almost all the time. We have experimentally verified that for small strings (<4 KiB) and less than 6 appends the cord is in fact flat [1].
- This change lifts the requirement of all ViewProxy types to be Copy. Our Cow types cannot be Copy because the owned types aren't copy.
[1] https://source.corp.google.com/piper///depot/google3/experimental/users/buchgr/cords/cords.cc
PiperOrigin-RevId: 655485943
`GPBTextFormatForMessage()` will contain any information for unknown fields.
And it clears the way for `GPBUnknownFieldSet` to eventually be deprecated
also.
PiperOrigin-RevId: 655184316
GenerateByteSize itself remains deeply nested, but by factoring out one part of
the loop, at least we make the part that generates UpdateByteSize a bit more
readable.
Making the callsite of MayEmitIfNonDefaultCheck less nested actually resulted
in slight readability improvements also in the generated code, namely of the
form:
@@ -10563,8 +10559,7 @@ PROTOBUF_NOINLINE void OneStringEdition:
{
// string data = 1;
- cached_has_bits =
- this_._impl_._has_bits_[0];
+ cached_has_bits = this_._impl_._has_bits_[0];
if (cached_has_bits & 0x00000001u) {
total_size += 1 + ::proto2::internal::WireFormatLite::StringSize(
this_._internal_data());
These readability improvements should be kept IMO -- they make the generated
protobuf C++ code slightly easier to read.
PiperOrigin-RevId: 655180880
The toolchain type is consumed by proto_library and produced by proto_toolchain rule. As such it's a private dependency, because both rules are now part of protobuf repo.
There are some early adopters of --incompatible_enable_proto_toolchain_resolution that might be broken from this: grpc, rules_go, rules_ts, rules_rust, rules_lint, because they have implementation that is not using proto_common. Those repositories need to define their own proto_lang_toolchain and consume it with proto_common.compile.
PiperOrigin-RevId: 654897871
Make a new internal api for collecting up the unknown group fields
that will be used in a future change.
Add testing for the new api and for unknown field parsing of groups in
general.
PiperOrigin-RevId: 654875605
When you have a `T: SomeTrait + ?Sized` and `trait SomeTrait:Sized`, the ?Sized has no effect (its not able to "remove" the requirement).
PiperOrigin-RevId: 654836513
Something has changed between Xcode and xcpretty such that
we don't get full logs when something fails to know what
file/line is actually failing; so stop using it.
PiperOrigin-RevId: 654830446
Also mark the helper as not inline able to avoid the code being over duplicated
within the file with some compiler optimizations turned on.
PiperOrigin-RevId: 654826102
ShouldEmitIfNonDefaultCheck is mostly called in MayEmitIfNonDefaultCheck,
except in one location in GenerateByteSize.
Making MayEmitIfNonDefaultCheck the only caller of ShouldEmitIfNonDefaultCheck
seems like a good application of DRY and reduces the possibility of unintended
divergence in the future. It also makes future changes to serialization logic
easier.
There should be no changes to the resulting generated code.
PiperOrigin-RevId: 654789393
This also changes MessageView and MessageMut to be subtraits of the ViewProxy and MutProxy (which also requires them to have lifetimes now, and a weird extra bounds on ViewProxy)
PiperOrigin-RevId: 654788207
- Now it is arena constructible. Allocates the array, strings, and inner sets in the arena.
- Now it is arena destructible. Will skip destructors when in an arena.
Also, optimize `default_instance()` now that we can provide a `constinit` version of it.
PiperOrigin-RevId: 654743861
Thomas Van Lenten
Protobuf Team Bot
Éamonn McManus
Adam Cozzette
Hong Shin
Protobuf Team Bot
Jakob Buchgraber
Nelson Liang
Derek Benson
Tony Liao
Chris Kennelly
Jie Luo
Evan Brown
Dmitri Gribenko