Protocol Buffers - Google's data interchange format (grpc依赖) https://developers.google.com/protocol-buffers/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

402 lines
16 KiB

// Protocol Buffers - Google's data interchange format
// Copyright 2008 Google Inc. All rights reserved.
// https://developers.google.com/protocol-buffers/
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
// * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
// This header is private to the ProtobolBuffers library and must NOT be
// included by any sources outside this library. The contents of this file are
// subject to change at any time without notice.
#import "GPBDescriptor.h"
#import "GPBWireFormat.h"
// Describes attributes of the field.
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
typedef NS_OPTIONS(uint16_t, GPBFieldFlags) {
GPBFieldNone = 0,
// These map to standard protobuf concepts.
GPBFieldRequired = 1 << 0,
GPBFieldRepeated = 1 << 1,
GPBFieldPacked = 1 << 2,
GPBFieldOptional = 1 << 3,
GPBFieldHasDefaultValue = 1 << 4,
// Indicate that the field should "clear" when set to zero value. This is the
// proto3 non optional behavior for singular data (ints, data, string, enum)
// fields.
GPBFieldClearHasIvarOnZero = 1 << 5,
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Indicates the field needs custom handling for the TextFormat name, if not
// set, the name can be derived from the ObjC name.
GPBFieldTextFormatNameCustom = 1 << 6,
// This flag has never had any meaning, it was set on all enum fields.
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
GPBFieldHasEnumDescriptor = 1 << 7,
// These are not standard protobuf concepts, they are specific to the
// Objective C runtime.
// These bits are used to mark the field as a map and what the key
// type is.
GPBFieldMapKeyMask = 0xF << 8,
GPBFieldMapKeyInt32 = 1 << 8,
GPBFieldMapKeyInt64 = 2 << 8,
GPBFieldMapKeyUInt32 = 3 << 8,
GPBFieldMapKeyUInt64 = 4 << 8,
GPBFieldMapKeySInt32 = 5 << 8,
GPBFieldMapKeySInt64 = 6 << 8,
GPBFieldMapKeyFixed32 = 7 << 8,
GPBFieldMapKeyFixed64 = 8 << 8,
GPBFieldMapKeySFixed32 = 9 << 8,
GPBFieldMapKeySFixed64 = 10 << 8,
GPBFieldMapKeyBool = 11 << 8,
GPBFieldMapKeyString = 12 << 8,
// If the enum for this field is "closed", meaning that it:
// - Has a fixed set of named values.
// - Encountering values not in this set causes them to be treated as unknown
// fields.
// - The first value (i.e., the default) may be nonzero.
// NOTE: This could be tracked just on the GPBEnumDescriptor, but to support
// previously generated code, there would be not data to get the behavior
// correct, so instead it is tracked on the field. If old source compatibility
// is removed, this could be removed and the GPBEnumDescription fetched from
// the GPBFieldDescriptor instead.
GPBFieldClosedEnum = 1 << 12,
};
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// NOTE: The structures defined here have their members ordered to minimize
// their size. This directly impacts the size of apps since these exist per
// field/extension.
typedef struct GPBFileDescription {
// The proto package for the file.
const char *package;
// The objc_class_prefix option if present.
const char *prefix;
// The file's proto syntax.
GPBFileSyntax syntax;
} GPBFileDescription;
// Describes a single field in a protobuf as it is represented as an ivar.
typedef struct GPBMessageFieldDescription {
// Name of ivar.
const char *name;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
union {
// className is deprecated and will be removed in favor of clazz.
// kept around right now for backwards compatibility.
// clazz is used iff GPBDescriptorInitializationFlag_UsesClassRefs is set.
char *className; // Name of the class of the message.
Class clazz; // Class of the message.
// For enums only.
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
GPBEnumDescriptorFunc enumDescFunc;
} dataTypeSpecific;
// The field number for the ivar.
uint32_t number;
// The index (in bits) into _has_storage_.
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// >= 0: the bit to use for a value being set.
// = GPBNoHasBit(INT32_MAX): no storage used.
// < 0: in a oneOf, use a full int32 to record the field active.
int32_t hasIndex;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Offset of the variable into it's structure struct.
uint32_t offset;
// Field flags. Use accessor functions below.
GPBFieldFlags flags;
// Data type of the ivar.
GPBDataType dataType;
} GPBMessageFieldDescription;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Fields in messages defined in a 'proto2' syntax file can provide a default
// value. This struct provides the default along with the field info.
typedef struct GPBMessageFieldDescriptionWithDefault {
// Default value for the ivar.
GPBGenericValue defaultValue;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
GPBMessageFieldDescription core;
} GPBMessageFieldDescriptionWithDefault;
// Describes attributes of the extension.
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
typedef NS_OPTIONS(uint8_t, GPBExtensionOptions) {
GPBExtensionNone = 0,
// These map to standard protobuf concepts.
GPBExtensionRepeated = 1 << 0,
GPBExtensionPacked = 1 << 1,
GPBExtensionSetWireFormat = 1 << 2,
};
// An extension
typedef struct GPBExtensionDescription {
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
GPBGenericValue defaultValue;
const char *singletonName;
// Before 3.12, `extendedClass` was just a `const char *`. Thanks to nested
// initialization
// (https://en.cppreference.com/w/c/language/struct_initialization#Nested_initialization) old
// generated code with `.extendedClass = GPBStringifySymbol(Something)` still works; and the
// current generator can use `extendedClass.clazz`, to pass a Class reference.
union {
const char *name;
Class clazz;
} extendedClass;
// Before 3.12, this was `const char *messageOrGroupClassName`. In the
// initial 3.12 release, we moved the `union messageOrGroupClass`, and failed
// to realize that would break existing source code for extensions. So to
// keep existing source code working, we added an unnamed union (C11) to
// provide both the old field name and the new union. This keeps both older
// and newer code working.
// Background: https://github.com/protocolbuffers/protobuf/issues/7555
union {
const char *messageOrGroupClassName;
union {
const char *name;
Class clazz;
} messageOrGroupClass;
};
GPBEnumDescriptorFunc enumDescriptorFunc;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
int32_t fieldNumber;
GPBDataType dataType;
GPBExtensionOptions options;
} GPBExtensionDescription;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
typedef NS_OPTIONS(uint32_t, GPBDescriptorInitializationFlags) {
GPBDescriptorInitializationFlag_None = 0,
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
GPBDescriptorInitializationFlag_FieldsWithDefault = 1 << 0,
GPBDescriptorInitializationFlag_WireFormat = 1 << 1,
// This is used as a stopgap as we move from using class names to class
// references. The runtime needs to support both until we allow a
// breaking change in the runtime.
GPBDescriptorInitializationFlag_UsesClassRefs = 1 << 2,
// This flag is used to indicate that the generated sources already contain
// the `GPBFieldClearHasIvarOnZero` flag and it doesn't have to be computed
// at startup. This allows older generated code to still work with the
// current runtime library.
GPBDescriptorInitializationFlag_Proto3OptionalKnown = 1 << 3,
// This flag is used to indicate that the generated sources already contain
// the `GPBFieldCloseEnum` flag and it doesn't have to be computed at startup.
// This allows the older generated code to still work with the current runtime
// library.
GPBDescriptorInitializationFlag_ClosedEnumSupportKnown = 1 << 4,
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
};
@interface GPBDescriptor () {
@package
NSArray *fields_;
NSArray *oneofs_;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
uint32_t storageSize_;
}
// fieldDescriptions and fileDescription have to be long lived, they are held as raw pointers.
+ (instancetype)allocDescriptorForClass:(Class)messageClass
messageName:(NSString *)messageName
fileDescription:(GPBFileDescription *)fileDescription
fields:(void *)fieldDescriptions
fieldCount:(uint32_t)fieldCount
storageSize:(uint32_t)storageSize
flags:(GPBDescriptorInitializationFlags)flags;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Called right after init to provide extra information to avoid init having
// an explosion of args. These pointers are recorded, so they are expected
// to live for the lifetime of the app.
- (void)setupOneofs:(const char **)oneofNames
count:(uint32_t)count
firstHasIndex:(int32_t)firstHasIndex;
- (void)setupExtraTextInfo:(const char *)extraTextFormatInfo;
- (void)setupExtensionRanges:(const GPBExtensionRange *)ranges count:(int32_t)count;
- (void)setupContainingMessageClass:(Class)msgClass;
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Deprecated, these remain to support older versions of source generation.
+ (instancetype)allocDescriptorForClass:(Class)messageClass
file:(GPBFileDescriptor *)file
fields:(void *)fieldDescriptions
fieldCount:(uint32_t)fieldCount
storageSize:(uint32_t)storageSize
flags:(GPBDescriptorInitializationFlags)flags;
+ (instancetype)allocDescriptorForClass:(Class)messageClass
rootClass:(Class)rootClass
file:(GPBFileDescriptor *)file
fields:(void *)fieldDescriptions
fieldCount:(uint32_t)fieldCount
storageSize:(uint32_t)storageSize
flags:(GPBDescriptorInitializationFlags)flags;
- (void)setupContainingMessageClassName:(const char *)msgClassName;
- (void)setupMessageClassNameSuffix:(NSString *)suffix;
@end
@interface GPBFileDescriptor ()
- (instancetype)initWithPackage:(NSString *)package
objcPrefix:(NSString *)objcPrefix
syntax:(GPBFileSyntax)syntax;
- (instancetype)initWithPackage:(NSString *)package syntax:(GPBFileSyntax)syntax;
@end
@interface GPBOneofDescriptor () {
@package
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
const char *name_;
NSArray *fields_;
SEL caseSel_;
}
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// name must be long lived.
- (instancetype)initWithName:(const char *)name fields:(NSArray *)fields;
@end
@interface GPBFieldDescriptor () {
@package
GPBMessageFieldDescription *description_;
GPB_UNSAFE_UNRETAINED GPBOneofDescriptor *containingOneof_;
SEL getSel_;
SEL setSel_;
SEL hasOrCountSel_; // *Count for map<>/repeated fields, has* otherwise.
SEL setHasSel_;
}
@end
typedef NS_OPTIONS(uint32_t, GPBEnumDescriptorInitializationFlags) {
GPBEnumDescriptorInitializationFlag_None = 0,
// Available: 1 << 0
// Marks this enum as a closed enum.
GPBEnumDescriptorInitializationFlag_IsClosed = 1 << 1,
};
@interface GPBEnumDescriptor ()
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// valueNames, values and extraTextFormatInfo have to be long lived, they are
// held as raw pointers.
+ (instancetype)allocDescriptorForName:(NSString *)name
valueNames:(const char *)valueNames
values:(const int32_t *)values
count:(uint32_t)valueCount
enumVerifier:(GPBEnumValidationFunc)enumVerifier
flags:(GPBEnumDescriptorInitializationFlags)flags;
+ (instancetype)allocDescriptorForName:(NSString *)name
valueNames:(const char *)valueNames
values:(const int32_t *)values
count:(uint32_t)valueCount
enumVerifier:(GPBEnumValidationFunc)enumVerifier
flags:(GPBEnumDescriptorInitializationFlags)flags
extraTextFormatInfo:(const char *)extraTextFormatInfo;
// Deprecated, these remain to support older versions of source generation.
+ (instancetype)allocDescriptorForName:(NSString *)name
valueNames:(const char *)valueNames
values:(const int32_t *)values
count:(uint32_t)valueCount
enumVerifier:(GPBEnumValidationFunc)enumVerifier;
+ (instancetype)allocDescriptorForName:(NSString *)name
valueNames:(const char *)valueNames
values:(const int32_t *)values
count:(uint32_t)valueCount
enumVerifier:(GPBEnumValidationFunc)enumVerifier
extraTextFormatInfo:(const char *)extraTextFormatInfo;
@end
@interface GPBExtensionDescriptor () {
@package
GPBExtensionDescription *description_;
}
@property(nonatomic, readonly) GPBWireFormat wireType;
// For repeated extensions, alternateWireType is the wireType with the opposite
// value for the packable property. i.e. - if the extension was marked packed
// it would be the wire type for unpacked; if the extension was marked unpacked,
// it would be the wire type for packed.
@property(nonatomic, readonly) GPBWireFormat alternateWireType;
// description has to be long lived, it is held as a raw pointer.
- (instancetype)initWithExtensionDescription:(GPBExtensionDescription *)desc
usesClassRefs:(BOOL)usesClassRefs;
// Deprecated. Calls above with `usesClassRefs = NO`
- (instancetype)initWithExtensionDescription:(GPBExtensionDescription *)desc;
- (NSComparisonResult)compareByFieldNumber:(GPBExtensionDescriptor *)other;
@end
CF_EXTERN_C_BEGIN
// Direct access is use for speed, to avoid even internally declaring things
// read/write, etc. The warning is enabled in the project to ensure code calling
// protos can turn on -Wdirect-ivar-access without issues.
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wdirect-ivar-access"
GPB_INLINE BOOL GPBFieldIsMapOrArray(GPBFieldDescriptor *field) {
return (field->description_->flags & (GPBFieldRepeated | GPBFieldMapKeyMask)) != 0;
}
GPB_INLINE GPBDataType GPBGetFieldDataType(GPBFieldDescriptor *field) {
return field->description_->dataType;
}
GPB_INLINE int32_t GPBFieldHasIndex(GPBFieldDescriptor *field) {
return field->description_->hasIndex;
}
GPB_INLINE uint32_t GPBFieldNumber(GPBFieldDescriptor *field) {
return field->description_->number;
}
GPB_INLINE BOOL GPBFieldIsClosedEnum(GPBFieldDescriptor *field) {
return (field->description_->flags & GPBFieldClosedEnum) != 0;
}
#pragma clang diagnostic pop
uint32_t GPBFieldTag(GPBFieldDescriptor *self);
// For repeated fields, alternateWireType is the wireType with the opposite
// value for the packable property. i.e. - if the field was marked packed it
// would be the wire type for unpacked; if the field was marked unpacked, it
// would be the wire type for packed.
uint32_t GPBFieldAlternateTag(GPBFieldDescriptor *self);
GPB_INLINE BOOL GPBExtensionIsRepeated(GPBExtensionDescription *description) {
return (description->options & GPBExtensionRepeated) != 0;
}
GPB_INLINE BOOL GPBExtensionIsPacked(GPBExtensionDescription *description) {
return (description->options & GPBExtensionPacked) != 0;
}
GPB_INLINE BOOL GPBExtensionIsWireFormat(GPBExtensionDescription *description) {
return (description->options & GPBExtensionSetWireFormat) != 0;
}
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Helper for compile time assets.
#ifndef GPBInternalCompileAssert
#define GPBInternalCompileAssert(test, msg) _Static_assert((test), #msg)
#endif // GPBInternalCompileAssert
Shrink ObjC overhead (generated size and some runtime sizes) NOTE: This is a binary breaking change as structure sizes have changed size and/or order. - Drop capturing field options, no other options were captured and other mobile targeted languages don't try to capture this sort information (saved 8 bytes for every field defined (in static data and again in field descriptor instance size data). - No longer generate/compile in the messages/enums in descriptor.proto. If developers need it, they should generate it and compile it in. Reduced the overhead of the core library. - Compute the number of has_bits actually needs to avoid over reserving. - Let the boolean single fields store via a has_bit to avoid storage, makes the common cases of the instance size smaller. - Reorder some flags and down size the enums to contain the bits needed. - Reorder the items in the structures to manually ensure they are are packed better (especially when generating 64bit code - 8 bytes for every field, 16 bytes for every extension, instance sizes 8 bytes also). - Split off the structure initialization so when the default is zero, the generated static storage doesn't need to reserve the space. This is batched at the message level, so all the fields for the message have to have zero defaults to get the saves. By definition all proto3 syntax files fall into this case but it also saves space for the proto2 that use the standard defaults. (saves 8 bytes of static data for every field that had a zero default) - Don't track the enums defined by a message. Nothing in the runtime needs it and it was just generation and runtime overhead. (saves 8 bytes per enum) - Ensure EnumDescriptors are started up threadsafe in all cases. - Split some of the Descriptor initialization into multiple methods so the generated code isn't padded with lots of zero/nil args. - Change how oneof info is feed to the runtime enabling us to generate less static data (8 bytes saved per oneof for 64bit). - Change how enum value informat is capture to pack the data and only decode it if it ends up being needed. Avoids padding issues causing bloat of 64bit, and removes the needs for extra pointers in addition to the data (just the data and one pointer now).
9 years ago
// Sanity check that there isn't padding between the field description
// structures with and without a default.
GPBInternalCompileAssert(sizeof(GPBMessageFieldDescriptionWithDefault) ==
(sizeof(GPBGenericValue) + sizeof(GPBMessageFieldDescription)),
DescriptionsWithDefault_different_size_than_expected);
CF_EXTERN_C_END