• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
12019-06-28 version 3.9.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
2
3  C++
4  * Optimize and simplify implementation of RepeatedPtrFieldBase
5  * Don't create unnecessary unknown field sets.
6  * Remove branch from accessors to repeated field element array.
7  * Added delimited parse and serialize util.
8  * Reduce size by not emitting constants for fieldnumbers
9  * Fix a bug when comparing finite and infinite field values with explicit tolerances.
10  * TextFormat::Parser should use a custom Finder to look up extensions by number if one is provided.
11  * Add MessageLite::Utf8DebugString() to make MessageLite more compatible with Message.
12  * Fail fast for better performance in DescriptorPool::FindExtensionByNumber() if descriptor has no defined extensions.
13  * Adding the file name to help debug colliding extensions
14  * Added FieldDescriptor::PrintableNameForExtension() and DescriptorPool::FindExtensionByPrintableName().
15    The latter will replace Reflection::FindKnownExtensionByName().
16  * Replace NULL with nullptr
17  * Created a new Add method in repeated field that allows adding a range of elements all at once.
18  * Enabled enum name-to-value mapping functions for C++ lite
19  * Avoid dynamic initialization in descriptor.proto generated code
20  * Move stream functions to MessageLite from Message.
21  * Move all zero_copy_stream functionality to io_lite.
22  * Do not create array of matched fields for simple repeated fields
23  * Enabling silent mode by default to reduce make compilation noise. (#6237)
24
25  Java
26  * Expose TextFormat.Printer and make it configurable. Deprecate the static methods.
27  * Library for constructing google.protobuf.Struct and google.protobuf.Value
28  * Make OneofDescriptor extend GenericDescriptor.
29  * Expose streamingness of service methods from MethodDescriptor.
30  * Fix a bug where TextFormat fails to parse Any filed with > 1 embedded message sub-fields.
31  * Establish consistent JsonFormat behavior for nulls in oneofs, regardless of order.
32  * Update GSON version to 3.8.5. (#6268)
33  * Add `protobuf_java_lite` Bazel target. (#6177)
34
35  Python
36  * Change implementation of Name() for enums that allow aliases in proto2 in Python
37    to be in line with claims in C++ implementation (to return first value).
38  * Explicitly say what field cannot be set when the new value fails a type check.
39  * Duplicate register in descriptor pool will raise errors
40  * Add __slots__ to all well_known_types classes, custom attributes are not allowed anymore.
41  * text_format only present 8 valid digits for float fields by default
42
43  JavaScript
44  * Add Oneof enum to the list of goog.provide
45
46  PHP
47  * Make php message class final to avoid mocking. (#6277)
48  * Rename get/setXXXValue to get/setXXXWrapper. (#6295)
49
50  Ruby
51  * Remove to_hash methods. (#6166)
52
53
542019-04-29 version 3.8.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
55
56  C++
57  * Use std::atomic<int32> in case of myriad2 platform
58  * Always declare enums to be int-sized
59  * Added DebugString() and ShortDebugString() methods on MessageLite
60  * Specialized different parse loop control flows
61  * Make hasbits potentially in register. The or's start forming an obstacle because it's a read modify store on the same mem address on each iteration.
62  * Move to an internal MACRO for parser validity checks.
63  * Improve map parsing performance.
64  * Make MergePartialFromCodedStream non virtual. This allows direct calls, potential inlining and is also a code health improvement
65  * Add an overall limit to parse_context to prevent reading past it. This allows to remove a annoying level of indirection.
66  * Fix a mistake, we shouldn't verify map key/value strings for utf8 in opt mode for proto2.
67  * Further improvements to cut binary size.
68  * Prepare to make MergePartialFromCodedStream non-virtual.
69  * A report on some interesting behavior change in python (caused by b/27494216) made me realize there is a check that needs to be done in case the parse ended on a end group tag.
70  * Add a note of caution to the comments around skip in CodedOutputStream.
71  * Simplify end check.
72  * Add overload for ParseMessage for MessageLite/Message types. If the explicit type is not known inlining won't help de-virtualizing the virtual call.
73  * Reduce linker input. It turns out that ParseMessage is not inlined, producing  template instantiations that are used only once and save nothing but cost more.
74  * Improve the parser.
75  * [c++17] Changed proto2::RepeatedPtrField iterators to no longer derive from the deprecated std::iterator class.
76  * Change the default value of case_insensitive_enum_parsing to false for JsonStringToMessage.
77  * Add a warning if a field name doesn't match the style guide.
78  * Fix TextFormat not round-trip correctly when float value is max float.
79  * Added locationed info for some errors at compiler
80  * Python reserved keywords are now working with getattr()/setattr() for most descriptors.
81  * Added AllowUnknownField() in text_format
82  * Append '_' to C++ reserved keywords for message, enum, extension
83  * Fix MSVC warning C4244 in protobuf's parse_context.h.
84  * Updating Iterators to be compatible with C++17 in MSVC.
85  * Use capability annotation in mutex.h
86  * Fix "UndefinedBehaviorSanitizer: cfi-bad-type"
87  * CriticalSectionLock class as a lightweight replacement for std::mutex on Windows platforms.
88  * Removed vestigial wire_format_lite_inl.h
89
90  C#
91  * Added System.Memory dependency.
92
93  Java
94  * Make Java protoc code generator ignore optimize_for LITE_RUNTIME. Users should instead use the Java lite protoc plugin.
95  * Change Extension getMessageDefaultInstance() to return Message instead of MessageLite.
96  * Prevent malicious input streams from leaking buffers for ByteString or ByteBuffer parsing.
97  * Release new Javalite runtime.
98  * Show warning in case potential file name conflict.
99  * Allow Java reserved keywords to be used in extensions.
100  * Added setAllowUnknownFields() in text format
101  * Add memoization to ExtensionRegistryLite.getEmptyRegistry()
102  * Improve performance of CodedOutputStream.writeUInt32NoTag
103  * Add an optimized mismatch-finding algorithm to UnsafeUtil.
104  * When serializing uint32 varints, check that we have MAX_VARINT32_SIZE bytes left, not just MAX_VARINT_SIZE.
105  * Minor optimization to RopeByteString.PieceIterator
106
107  JavaScript
108  * Simplify generated toObject code when the default value is used.
109
110  Python
111  * Changes implementation of Name() for enums that allow aliases in proto2 in Python to be in line with claims in C++ implementation (to return first value).
112  * Added double_format option in text format printer.
113  * Added iter and __contains__ to extension dict
114  * Added allow_unknown_field option in python text format parser
115  * Fixed Timestamp.ToDatetime() loses precision issue
116  * Support unknown field in text format printer.
117  * Float field will be convert to inf if bigger than struct.unpack('f', b'\xff\xff\x7f\x7f')[0] which is about 3.4028234664e+38,
118  convert to -inf if smaller than -3.4028234664e+38
119  * Allowed casting str->bytes in Message.__setstate__
120
121  Ruby
122  * Helper methods to get enum name for Ruby.
123
124
1252019-01-24 version 3.7.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
126
127  C++
128  * Introduced new MOMI (maybe-outside-memory-interval) parser.
129  * Add an option to json_util to parse enum as case-insensitive. In the future, enum parsing in json_util will become case-sensitive.
130  * Added conformance test for enum aliases
131  * Added support for --cpp_out=speed:...
132  * Added use of C++ override keyword where appropriate
133  * Many other cleanups and fixes.
134
135  Java
136  * Fix illegal reflective access warning in JDK 9+
137  * Add BOM
138
139  Python
140  * Added Python 3.7 compatibility.
141  * Modified ParseFromString to return bytes parsed .
142  * Introduce Proto C API.
143  * FindFileContainingSymbol in descriptor pool is now able to find field and enum values.
144  * reflection.MakeClass()  and  reflection.ParseMessage() are deprecated.
145  * Added DescriptorPool.FindMethodByName() method in pure python (c extension alreay has it)
146  * Flipped proto3 to preserve unknown fields by default.
147  * Added support for memoryview in python3 proto message parsing.
148  * Added MergeFrom for repeated scalar fields in c extension (pure python already has it)
149  * Surrogates are now rejected at setters in python3.
150  * Added public unknown field API.
151  * RecursionLimit is also set to max if allow_oversize_protos is enabled.
152  * Disallow duplicate scalars in proto3 text_format parse.
153  * Fix some segment faults for c extension map field.
154
155  PHP
156  * Most issues for json encoding/decoding in the c extension have been fixed. There are still some edge cases not fixed. For more details, check conformance/failure_list_php_c.txt.
157  * Supports php 7.3
158  * Added helper methods to convert between enum values and names.
159  * Allow setting/getting wrapper message fields using primitive values.
160  * Various bug fixes.
161
162  Ruby
163  * Ruby 2.6 support.
164  * Drops support for ruby < 2.3.
165  * Most issues for json encoding/decoding in the c extension have been fixed. There are still some edge cases not fixed. For more details, check conformance/failure_list_ruby.txt.
166  * Json parsing can specify an option to ignore unknown fields: msg.decode_json(data, {ignore_unknown_fields: true}).
167  * Added support for proto2 syntax (partially).
168  * Various bug fixes.
169
170  Csharp
171  * More support for FieldMask include merge, intersect and more.
172  * Increasing the default recursion limit to 100.
173  * Support loading FileDescriptors dynamically.
174  * Provide access to comments from descriptors.
175  * Added Any.Is method.
176  * Compatible with C# 6
177  * Added IComparable and comparison operators on Timestamp.
178
179  Objective C
180  * Add ability to introspect list of enum values (#4678)
181  * Copy the value when setting message/data fields (#5215)
182  * Support suppressing the objc package prefix checks on a list of files (#5309)
183  * More complete keyword and NSObject method (via categories) checks for field names, can result in more fields being rename, but avoids the collisions at runtime (#5289)
184  * Small fixes to TextFormat generation for extensions (#5362)
185  * Provide more details/context in deprecation messages (#5412)
186  * Array/Dictionary enumeration blocks NS_NOESCAPE annotation for Swift (#5421)
187  * Properly annotate extensions for ARC when their names imply behaviors (#5427)
188  * Enum alias name collision improvements (#5480)
189
190
1912018-07-27 version 3.6.1 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
192
193  C++
194  * Introduced workaround for Windows issue with std::atomic and std::once_flag
195    initialization (#4777, #4773).
196
197  PHP
198  * Added compatibility with PHP 7.3 (#4898).
199
200  Ruby
201  * Fixed Ruby crash involving Any encoding (#4718).
202
2032018-06-01 version 3.6.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
204
205  C++
206  * Starting from this release, we now require C++11. For those we cannot yet
207    upgrade to C++11, we will try to keep the 3.5.x branch updated with
208    critical bug fixes only. If you have any concerns about this, please
209    comment on issue #2780.
210  * Moved to C++11 types like std::atomic and std::unique_ptr and away from our
211    old custom-built equivalents.
212  * Added support for repeated message fields in lite protos using implicit
213    weak fields. This is an experimental feature that allows the linker to
214    strip out more unused messages than previously was possible.
215  * Fixed SourceCodeInfo for interpreted options and extension range options.
216  * Fixed always_print_enums_as_ints option for JSON serialization.
217  * Added support for ignoring unknown enum values when parsing JSON.
218  * Create std::string in Arena memory.
219  * Fixed ValidateDateTime to correctly check the day.
220  * Fixed bug in ZeroCopyStreamByteSink.
221  * Various other cleanups and fixes.
222
223  Java
224  * Dropped support for Java 6.
225  * Added a UTF-8 decoder that uses Unsafe to directly decode a byte buffer.
226  * Added deprecation annotations to generated code for deprecated oneof
227    fields.
228  * Fixed map field serialization in DynamicMessage.
229  * Cleanup and documentation for Java Lite runtime.
230  * Various other fixes and cleanups
231  * Fixed unboxed arraylists to handle an edge case
232  * Improved performance for copying between unboxed arraylists
233  * Fixed lite protobuf to avoid Java compiler warnings
234  * Improved test coverage for lite runtime
235  * Performance improvements for lite runtime
236
237  Python
238  * Fixed bytes/string map key incompatibility between C++ and pure-Python
239    implementations (issue #4029)
240  * Added __init__.py files to compiler and util subpackages
241  * Use /MT for all Windows versions
242  * Fixed an issue affecting the Python-C++ implementation when used with
243    Cython (issue #2896)
244  * Various text format fixes
245  * Various fixes to resolve behavior differences between the pure-Python and
246    Python-C++ implementations
247
248  PHP
249  * Added php_metadata_namespace to control the file path of generated metadata
250    file.
251  * Changed generated classes of nested message/enum. E.g., Foo.Bar, which
252    previously generates Foo_Bar, now generates Foo/Bar
253  * Added array constructor. When creating a message, users can pass a php
254    array whose content is field name to value pairs into constructor. The
255    created message will be initialized according to the array. Note that
256    message field should use a message value instead of a sub-array.
257  * Various bug fixes.
258
259  Objective-C
260  * We removed some helper class methods from GPBDictionary to shrink the size
261    of the library, the functionary is still there, but you may need to do some
262    specific +alloc / -init… methods instead.
263  * Minor improvements in the performance of object field getters/setters by
264    avoiding some memory management overhead.
265  * Fix a memory leak during the raising of some errors.
266  * Make header importing completely order independent.
267  * Small code improvements for things the undefined behaviors compiler option
268    was flagging.
269
270  Ruby
271  * Added ruby_package file option to control the module of generated class.
272  * Various bug fixes.
273
274  Javascript
275  * Allow setting string to int64 field.
276
277  Csharp
278  * Unknown fields are now parsed and then sent back on the wire. They can be
279    discarded at parse time via a CodedInputStream option.
280  * Movement towards working with .NET 3.5 and Unity
281  * Expression trees are no longer used
282  * AOT generics issues in Unity/il2cpp have a workaround (see this commit for
283    details)
284  * Floating point values are now compared bitwise (affects NaN value
285    comparisons)
286  * The default size limit when parsing is now 2GB rather than 64MB
287  * MessageParser now supports parsing from a slice of a byte array
288  * JSON list parsing now accepts null values where the underlying proto
289    representation does
290
2912017-12-20 version 3.5.1 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
292  Planned Future Changes
293  * Make C++ implementation C++11 only: we plan to require C++11 to build
294    protobuf code starting from 3.6.0 release. Please join this github issue:
295    https://github.com/protocolbuffers/protobuf/issues/2780 to provide your feedback.
296
297  protoc
298  * Fixed a bug introduced in 3.5.0 and protoc in Windows now accepts non-ascii
299    characters in paths again.
300
301  C++
302  * Removed several usages of C++11 features in the code base.
303  * Fixed some compiler warnings.
304
305  PHP
306  * Fixed memory leak in C-extension implementation.
307  * Added discardUnknokwnFields API.
308  * Removed duplicatd typedef in C-extension headers.
309  * Avoided calling private php methods (timelib_update_ts).
310  * Fixed Any.php to use fully-qualified name for DescriptorPool.
311
312  Ruby
313  * Added Google_Protobuf_discard_unknown for discarding unknown fields in
314    messages.
315
316  C#
317  * Unknown fields are now preserved by default.
318  * Floating point values are now bitwise compared, affecting message equality
319    check and Contains() API in map and repeated fields.
320
321
3222017-11-13 version 3.5.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
323  Planned Future Changes
324  * Make C++ implementation C++11 only: we plan to require C++11 to build
325    protobuf code starting from 3.6.0 release. Please join this github issue:
326    https://github.com/protocolbuffers/protobuf/issues/2780 to provide your feedback.
327
328  General
329  * Unknown fields are now preserved in proto3 for most of the language
330    implementations for proto3 by default. See the per-language section for
331    details.
332  * reserve keyword are now supported in enums
333
334  C++
335  * Proto3 messages are now preserving unknown fields by default. If you rely on
336    unknowns fields being dropped. Please use DiscardUnknownFields() explicitly.
337  * Deprecated the unsafe_arena_release_* and unsafe_arena_add_allocated_*
338    methods for string fields.
339  * Added move constructor and move assignment to RepeatedField,
340    RepeatedPtrField and google::protobuf::Any.
341  * Added perfect forwarding in Arena::CreateMessage
342  * In-progress experimental support for implicit weak fields with lite protos.
343    This feature allows the linker to strip out more unused messages and reduce
344    binary size.
345  * Various performance optimizations.
346
347  Java
348  * Proto3 messages are now preserving unknown fields by default. If you’d like
349    to drop unknown fields, please use the DiscardUnknownFieldsParser API. For
350    example:
351      Parser<Foo> parser = DiscardUnknownFieldsParser.wrap(Foo.parser());
352      Foo foo = parser.parseFrom(input);
353  * Added a new CodedInputStream decoder for Iterable<ByteBuffer> with direct
354    ByteBuffers.
355  * TextFormat now prints unknown length-delimited fields as messages if
356    possible.
357  * FieldMaskUtil.merge() no longer creates unnecessary empty messages when a
358    message field is unset in both source message and destination message.
359  * Various performance optimizations.
360
361  Python
362  * Proto3 messages are now preserving unknown fields by default. Use
363    message.DiscardUnknownFields() to drop unknown fields.
364  * Add FieldDescriptor.file in generated code.
365  * Add descriptor pool FindOneofByName in pure python.
366  * Change unknown enum values into unknown field set .
367  * Add more Python dict/list compatibility for Struct/ListValue.
368  * Add utf-8 support for text_format.Merge()/Parse().
369  * Support numeric unknown enum values for proto3 JSON format.
370  * Add warning for Unexpected end-group tag in cpp extension.
371
372  PHP
373  * Proto3 messages are now preserving unknown fields.
374  * Provide well known type messages in runtime.
375  * Add prefix ‘PB’ to generated class of reserved names.
376  * Fixed all conformance tests for encode/decode json in php runtime. C
377    extension needs more work.
378
379  Objective-C
380  * Fixed some issues around copying of messages with unknown fields and then
381    mutating the unknown fields in the copy.
382
383  C#
384  * Added unknown field support in JsonParser.
385  * Fixed oneof message field merge.
386  * Simplify parsing messages from array slices.
387
388  Ruby
389  * Unknown fields are now preserved by default.
390  * Fixed several bugs for segment fault.
391
392  Javascript
393  * Decoder can handle both paced and unpacked data no matter how the proto is
394    defined.
395  * Decoder now accept long varint for 32 bit integers.
396
397
3982017-08-14 version 3.4.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
399  Planned Future Changes
400  * There are some changes that are not included in this release but are planned
401    for the near future
402      - Preserve unknown fields in proto3: We are going to bring unknown fields
403        back into proto3. In this release, some languages start to support
404        preserving unknown fields in proto3, controlled by flags/options. Some
405        languages also introduce explicit APIs to drop unknown fields for
406        migration. Please read the change log sections by languages for details.
407        For general timeline and plan:
408
409          https://docs.google.com/document/d/1KMRX-G91Aa-Y2FkEaHeeviLRRNblgIahbsk4wA14gRk/view
410
411        For issues and discussions:
412
413          https://github.com/protocolbuffers/protobuf/issues/272
414
415      - Make C++ implementation C++11 only: we plan to require C++11 to build
416        protobuf code starting from 3.5.0 or 3.6.0 release, after unknown fields
417        semantic changes are finished. Please join this
418        github issue:
419
420          https://github.com/protocolbuffers/protobuf/issues/2780
421
422        to provide your feedback.
423
424  General
425  * Extension ranges now accept options and are customizable.
426  * "reserve" keyword now supports “max” in field number ranges,
427    e.g.  reserve 1000 to max;
428
429  C++
430  * Proto3 messages are now able to preserve unknown fields. The default
431    behavior is still to drop unknowns, which will be flipped in a future
432    release. If you rely on unknowns fields being dropped. Please use
433    Message::DiscardUnknownFields() explicitly.
434  * Packable proto3 fields are now packed by default in serialization.
435  * Following C++11 features are introduced when C++11 is available:
436      - move-constructor and move-assignment are introduced to messages
437      - Repeated fields constructor now takes std::initializer_list
438      - rvalue setters are introduced for string fields
439  * Experimental Table-Driven parsing and serialization available to test. To
440    enable it, pass in table_driven_parsing table_driven_serialization protoc
441    generator flags for C++
442
443      $ protoc --cpp_out=table_driven_parsing,table_driven_serialization:./ \
444        test.proto
445
446  * lite generator parameter supported by the generator. Once set, all generated
447    files, use lite runtime regardless of the optimizer_for setting in the
448    .proto file.
449  * Various optimizations to make C++ code more performant on PowerPC platform
450  * Fixed maps data corruption when the maps are modified by both reflection API
451    and generated API.
452  * Deterministic serialization on maps reflection now uses stable sort.
453  * file() accessors are introduced to various *Descriptor classes to make
454    writing template function easier.
455  * ByteSize() and SpaceUsed() are deprecated.Use ByteSizeLong() and
456    SpaceUsedLong() instead
457  * Consistent hash function is used for maps in DEBUG and NDEBUG build.
458  * "using namespace std" is removed from stubs/common.h
459  * Various performance optimizations and bug fixes
460
461  Java
462  * Introduced new parser API DiscardUnknownFieldsParser in preparation of
463    proto3 unknown fields preservation change. Users who want to drop unknown
464    fields should migrate to use this new parser API. For example:
465
466      Parser<Foo> parser = DiscardUnknownFieldsParser.wrap(Foo.parser());
467      Foo foo = parser.parseFrom(input);
468
469  * Introduced new TextFormat API printUnicodeFieldValue() that prints field
470    value without escaping unicode characters.
471  * Added Durations.compare(Duration, Duration) and
472    Timestamps.compare(Timestamp, Timestamp).
473  * JsonFormat now accepts base64url encoded bytes fields.
474  * Optimized CodedInputStream to do less copies when parsing large bytes
475    fields.
476  * Optimized TextFormat to allocate less memory when printing.
477
478  Python
479  * SerializeToString API is changed to SerializeToString(self, **kwargs),
480    deterministic parameter is accepted for deterministic serialization.
481  * Added sort_keys parameter in json format to make the output deterministic.
482  * Added indent parameter in json format.
483  * Added extension support in json format.
484  * Added __repr__ support for repeated field in cpp implementation.
485  * Added file in FieldDescriptor.
486  * Added pretty-print filter to text format.
487  * Services and method descriptors are always printed even if generic_service
488    option is turned off.
489  * Note: AppEngine 2.5 is deprecated on June 2017 that AppEngine 2.5 will
490    never update protobuf runtime. Users who depend on AppEngine 2.5 should use
491    old protoc.
492
493  PHP
494  * Support PHP generic services. Specify file option php_generic_service=true
495    to enable generating service interface.
496  * Message, repeated and map fields setters take value instead of reference.
497  * Added map iterator in c extension.
498  * Support json  encode/decode.
499  * Added more type info in getter/setter phpdoc
500  * Fixed the problem that c extension and php implementation cannot be used
501    together.
502  * Added file option php_namespace to use custom php namespace instead of
503    package.
504  * Added fluent setter.
505  * Added descriptor API in runtime for custom encode/decode.
506  * Various bug fixes.
507
508  Objective-C
509  * Fix for GPBExtensionRegistry copying and add tests.
510  * Optimize GPBDictionary.m codegen to reduce size of overall library by 46K
511    per architecture.
512  * Fix some cases of reading of 64bit map values.
513  * Properly error on a tag with field number zero.
514  * Preserve unknown fields in proto3 syntax files.
515  * Document the exceptions on some of the writing apis.
516
517  C#
518  * Implemented IReadOnlyDictionary<K,V> in MapField<K,V>
519  * Added TryUnpack method for Any message in addition to Unpack.
520  * Converted C# projects to MSBuild (csproj) format.
521
522  Ruby
523  * Several bug fixes.
524
525  Javascript
526  * Added support of field option js_type. Now one can specify the JS type of a
527    64-bit integer field to be string in the generated code by adding option
528    [jstype = JS_STRING] on the field.
529
5302017-04-05 version 3.3.0 (C++/Java/Python/PHP/Objective-C/C#/Ruby/JavaScript)
531  Planned Future Changes
532  * There are some changes that are not included in this release but are
533    planned for the near future:
534      - Preserve unknown fields in proto3: please read this doc:
535
536          https://docs.google.com/document/d/1KMRX-G91Aa-Y2FkEaHeeviLRRNblgIahbsk4wA14gRk/view
537
538        for the timeline and follow up this github issue:
539
540          https://github.com/protocolbuffers/protobuf/issues/272
541
542        for discussion.
543      - Make C++ implementation C++11 only: we plan to require C++11 to build
544        protobuf code starting from 3.4.0 or 3.5.0 release. Please join this
545        github issue:
546
547          https://github.com/protocolbuffers/protobuf/issues/2780
548
549        to provide your feedback.
550
551  C++
552  * Fixed map fields serialization of DynamicMessage to correctly serialize
553    both key and value regardless of their presence.
554  * Parser now rejects field number 0 correctly.
555  * New API Message::SpaceUsedLong() that’s equivalent to
556    Message::SpaceUsed() but returns the value in size_t.
557  * JSON support
558    - New flag always_print_enums_as_ints in JsonPrintOptions.
559    - New flag preserve_proto_field_names in JsonPrintOptions. It will instruct
560      the JSON printer to use the original field name declared in the .proto
561      file instead of converting them to lowerCamelCase when printing JSON.
562    - JsonPrintOptions.always_print_primtive_fields now works for oneof message
563      fields.
564    - Fixed a bug that doesn’t allow different fields to set the same json_name
565      value.
566    - Fixed a performance bug that causes excessive memory copy when printing
567      large messages.
568  * Various performance optimizations.
569
570  Java
571  * Map field setters eagerly validate inputs and throw NullPointerExceptions
572    as appropriate.
573  * Added ByteBuffer overloads to the generated parsing methods and the Parser
574    interface.
575  * proto3 enum's getNumber() method now throws on UNRECOGNIZED values.
576  * Output of JsonFormat is now locale independent.
577
578  Python
579  * Added FindServiceByName() in the pure-Python DescriptorPool. This works only
580    for descriptors added with DescriptorPool.Add(). Generated descriptor_pool
581    does not support this yet.
582  * Added a descriptor_pool parameter for parsing Any in text_format.Parse().
583  * descriptor_pool.FindFileContainingSymbol() now is able to find nested
584    extensions.
585  * Extending empty [] to repeated field now sets parent message presence.
586
587  PHP
588  * Added file option php_class_prefix. The prefix will be prepended to all
589    generated classes defined in the file.
590  * When encoding, negative int32 values are sign-extended to int64.
591  * Repeated/Map field setter accepts a regular PHP array. Type checking is
592    done on the array elements.
593  * encode/decode are renamed to serializeToString/mergeFromString.
594  * Added mergeFrom, clear method on Message.
595  * Fixed a bug that oneof accessor didn’t return the field name that is
596    actually set.
597  * C extension now works with php7.
598  * This is the first GA release of PHP. We guarantee that old generated code
599    can always work with new runtime and new generated code.
600
601  Objective-C
602  * Fixed help for GPBTimestamp for dates before the epoch that contain
603    fractional seconds.
604  * Added GPBMessageDropUnknownFieldsRecursively() to remove unknowns from a
605    message and any sub messages.
606  * Addressed a threading race in extension registration/lookup.
607  * Increased the max message parsing depth to 100 to match the other languages.
608  * Removed some use of dispatch_once in favor of atomic compare/set since it
609    needs to be heap based.
610  * Fixes for new Xcode 8.3 warnings.
611
612  C#
613  * Fixed MapField.Values.CopyTo, which would throw an exception unnecessarily
614    if provided exactly the right size of array to copy to.
615  * Fixed enum JSON formatting when multiple names mapped to the same numeric
616    value.
617  * Added JSON formatting option to format enums as integers.
618  * Modified RepeatedField<T> to implement IReadOnlyList<T>.
619  * Introduced the start of custom option handling; it's not as pleasant as it
620    might be, but the information is at least present. We expect to extend code
621    generation to improve this in the future.
622  * Introduced ByteString.FromStream and ByteString.FromStreamAsync to
623    efficiently create a ByteString from a stream.
624  * Added whole-message deprecation, which decorates the class with [Obsolete].
625
626  Ruby
627  * Fixed Message#to_h for messages with map fields.
628  * Fixed memcpy() in binary gems to work for old glibc, without breaking the
629    build for non-glibc libc’s like musl.
630
631  Javascript
632  * Added compatibility tests for version 3.0.0.
633  * Added conformance tests.
634  * Fixed serialization of extensions: we need to emit a value even if it is
635    falsy (like the number 0).
636  * Use closurebuilder.py in favor of calcdeps.py for compiling JavaScript.
637
6382017-01-23 version 3.2.0 (C++/Java/Python/PHP/Ruby/Objective-C/C#/JavaScript/Lite)
639  General
640  * Added protoc version number to protoc plugin protocol. It can be used by
641    protoc plugin to detect which version of protoc is used with the plugin and
642    mitigate known problems in certain version of protoc.
643
644  C++
645  * The default parsing byte size limit has been raised from 64MB to 2GB.
646  * Added rvalue setters for non-arena string fields.
647  * Enabled debug logging for Android.
648  * Fixed a double-free problem when using Reflection::SetAllocatedMessage()
649    with extension fields.
650  * Fixed several deterministic serialization bugs:
651    * MessageLite::SerializeAsString() now respects the global deterministic
652      serialization flag.
653    * Extension fields are serialized deterministically as well.  Fixed protocol
654      compiler to correctly report importing-self as an error.
655  * Fixed FileDescriptor::DebugString() to print custom options correctly.
656  * Various performance/codesize optimizations and cleanups.
657
658  Java
659  * The default parsing byte size limit has been raised from 64MB to 2GB.
660  * Added recursion limit when parsing JSON.
661  * Fixed a bug that enumType.getDescriptor().getOptions() doesn't have custom
662    options.
663  * Fixed generated code to support field numbers up to 2^29-1.
664
665  Python
666  * You can now assign NumPy scalars/arrays (np.int32, np.int64) to protobuf
667    fields, and assigning other numeric types has been optimized for
668    performance.
669  * Pure-Python: message types are now garbage-collectable.
670  * Python/C++: a lot of internal cleanup/refactoring.
671
672  PHP (Alpha)
673  * For 64-bit integers type (int64/uint64/sfixed64/fixed64/sint64), use PHP
674    integer on 64-bit environment and PHP string on 32-bit environment.
675  * PHP generated code also conforms to PSR-4 now.
676  * Fixed ZTS build for c extension.
677  * Fixed c extension build on Mac.
678  * Fixed c extension build on 32-bit linux.
679  * Fixed the bug that message without namespace is not found in the descriptor
680    pool. (#2240)
681  * Fixed the bug that repeated field is not iterable in c extension.
682  * Message names Empty will be converted to GPBEmpty in generated code.
683  * Added phpdoc in generated files.
684  * The released API is almost stable. Unless there is large problem, we won't
685    change it. See
686    https://developers.google.com/protocol-buffers/docs/reference/php-generated
687    for more details.
688
689  Objective-C
690  * Added support for push/pop of the stream limit on CodedInputStream for
691    anyone doing manual parsing.
692
693  C#
694  * No changes.
695
696  Ruby
697  * Message objects now support #respond_to? for field getters/setters.
698  * You can now compare “message == non_message_object” and it will return false
699    instead of throwing an exception.
700  * JRuby: fixed #hashCode to properly reflect the values in the message.
701
702  Javascript
703  * Deserialization of repeated fields no longer has quadratic performance
704    behavior.
705  * UTF-8 encoding/decoding now properly supports high codepoints.
706  * Added convenience methods for some well-known types: Any, Struct, and
707    Timestamp. These make it easier to convert data between native JavaScript
708    types and the well-known protobuf types.
709
7102016-09-23 version 3.1.0 (C++/Java/Python/PHP/Ruby/Objective-C/C#/JavaScript/Lite)
711  General
712  * Proto3 support in PHP (alpha).
713  * Various bug fixes.
714
715  C++
716  * Added MessageLite::ByteSizeLong() that’s equivalent to
717    MessageLite::ByteSize() but returns the value in size_t. Useful to check
718    whether a message is over the 2G size limit that protobuf can support.
719  * Moved default_instances to global variables. This allows default_instance
720    addresses to be known at compile time.
721  * Adding missing generic gcc 64-bit atomicops.
722  * Restore New*Callback into google::protobuf namespace since these are used
723    by the service stubs code
724  * JSON support.
725    * Fixed some conformance issues.
726  * Fixed a JSON serialization bug for bytes fields.
727
728  Java
729  * Fixed a bug in TextFormat that doesn’t accept empty repeated fields (i.e.,
730    “field: [ ]”).
731  * JSON support
732    * Fixed JsonFormat to do correct snake_case-to-camelCase conversion for
733      non-style-conforming field names.
734    * Fixed JsonFormat to parse empty Any message correctly.
735    * Added an option to JsonFormat.Parser to ignore unknown fields.
736  * Experimental API
737    * Added UnsafeByteOperations.unsafeWrap(byte[]) to wrap a byte array into
738      ByteString without copy.
739
740  Python
741  * JSON support
742    * Fixed some conformance issues.
743
744  PHP (Alpha)
745  * We have added the proto3 support for PHP via both a pure PHP package and a
746    native c extension. The pure PHP package is intended to provide usability
747    to wider range of PHP platforms, while the c extension is intended to
748    provide higher performance. Both implementations provide the same runtime
749    APIs and share the same generated code. Users don’t need to re-generate
750    code for the same proto definition when they want to switch the
751    implementation later. The pure PHP package is included in the php/src
752    directory, and the c extension is included in the php/ext directory.
753
754    Both implementations provide idiomatic PHP APIs:
755    * All messages and enums are defined as PHP classes.
756    * All message fields can only be accessed via getter/setter.
757    * Both repeated field elements and map elements are stored in containers
758      that act like a normal PHP array.
759
760    Unlike several existing third-party PHP implementations for protobuf, our
761    implementations are built on a "strongly-typed" philosophy: message fields
762    and array/map containers will throw exceptions eagerly when values of the
763    incorrect type (not including those that can be type converted, e.g.,
764    double <-> integer <-> numeric string) are inserted.
765
766    Currently, pure PHP runtime supports php5.5, 5.6 and 7 on linux. C
767    extension runtime supports php5.5 and 5.6 on linux.
768
769    See php/README.md for more details about installment. See
770    https://developers.google.com/protocol-buffers/docs/phptutorial for more
771    details about APIs.
772
773  Objective-C
774  * Helpers are now provided for working the Any well known type (see
775    GPBWellKnownTypes.h for the api additions).
776  * Some improvements in startup code (especially when extensions aren’t used).
777
778  Javascript
779  * Fixed missing import of jspb.Map
780  * Fixed valueWriterFn variable name
781
782  Ruby
783  * Fixed hash computation for JRuby's RubyMessage
784  * Make sure map parsing frames are GC-rooted.
785  * Added API support for well-known types.
786
787  C#
788  * Removed check on dependency in the C# reflection API.
789
7902016-09-06 version 3.0.2 (C++/Java/Python/Ruby/Objective-C/C#/JavaScript/Lite)
791  General
792  * Various bug fixes.
793
794  Objective C
795  * Fix for oneofs in proto3 syntax files where fields were set to the zero
796    value.
797  * Fix for embedded null character in strings.
798  * CocoaDocs support
799
800  Ruby
801  * Fixed memory corruption bug in parsing that could occur under GC pressure.
802
803  Javascript
804  * jspb.Map is now properly exported to CommonJS modules.
805
806  C#
807  * Removed legacy_enum_values flag.
808
809
8102016-07-27 version 3.0.0 (C++/Java/Python/Ruby/Objective-C/C#/JavaScript/Lite)
811  General
812  * This log only contains changes since the beta-4 release. Summarized change
813    log since the last stable release (v2.6.1) can be found in the github
814    release page.
815
816  Compatibility Notice
817  * v3.0.0 is the first API stable release of the v3.x series. We do not expect
818    any future API breaking changes.
819  * For C++, Java Lite and Objective-C, source level compatibility is
820    guaranteed.  Upgrading from v3.0.0 to newer minor version releases will be
821    source compatible. For example, if your code compiles against protobuf
822    v3.0.0, it will continue to compile after you upgrade protobuf library to
823    v3.1.0.
824  * For other languages, both source level compatibility and binary level
825    compatibility are guaranteed. For example, if you have a Java binary built
826    against protobuf v3.0.0. After switching the protobuf runtime binary to
827    v3.1.0, your built binary should continue to work.
828  * Compatibility is only guaranteed for documented API and documented
829    behaviors. If you are using undocumented API (e.g., use anything in the C++
830    internal namespace), it can be broken by minor version releases in an
831    undetermined manner.
832
833  Ruby
834  * When you assign a string field `a.string_field = "X"`, we now call
835    #encode(UTF-8) on the string and freeze the copy. This saves you from
836    needing to ensure the string is already encoded as UTF-8. It also prevents
837    you from mutating the string after it has been assigned (this is how we
838    ensure it stays valid UTF-8).
839  * The generated file for `foo.proto` is now `foo_pb.rb` instead of just
840    `foo.rb`. This makes it easier to see which imports/requires are from
841    protobuf generated code, and also prevents conflicts with any `foo.rb` file
842    you might have written directly in Ruby. It is a backward-incompatible
843    change: you will need to update all of your `require` statements.
844  * For package names like `foo_bar`, we now translate this to the Ruby module
845    `FooBar`. This is more idiomatic Ruby than what we used to do (`Foo_bar`).
846
847  JavaScript
848  * Scalar fields like numbers and boolean now return defaults instead of
849    `undefined` or `null` when they are unset. You can test for presence
850    explicitly by calling `hasFoo()`, which we now generate for scalar fields.
851
852  Java Lite
853  * Java Lite is now implemented as a separate plugin, maintained in the
854    `javalite` branch. Both lite runtime and protoc artifacts will be available
855    in Maven.
856
857  C#
858  * Target platforms now .NET 4.5, selected portable subsets and .NET Core.
859  * legacy_enum_values option is no longer supported.
860
8612016-07-15 version 3.0.0-beta-4 (C++/Java/Python/Ruby/Objective-C/C#/JavaScript)
862  General
863  * Added a deterministic serialization API for C++. The deterministic
864    serialization guarantees that given a binary, equal messages will be
865    serialized to the same bytes. This allows applications like MapReduce to
866    group equal messages based on the serialized bytes. The deterministic
867    serialization is, however, NOT canonical across languages; it is also
868    unstable across different builds with schema changes due to unknown fields.
869    Users who need canonical serialization, e.g. persistent storage in a
870    canonical form, fingerprinting, etc, should define their own
871    canonicalization specification and implement the serializer using reflection
872    APIs rather than relying on this API.
873  * Added OneofOptions. You can now define custom options for oneof groups.
874      import "google/protobuf/descriptor.proto";
875      extend google.protobuf.OneofOptions {
876        optional int32 my_oneof_extension = 12345;
877      }
878      message Foo {
879        oneof oneof_group {
880          (my_oneof_extension) = 54321;
881          ...
882        }
883      }
884
885  C++ (beta)
886  * Introduced a deterministic serialization API in
887    CodedOutputStream::SetSerializationDeterministic(bool). See the notes about
888    deterministic serialization in the General section.
889  * Added google::protobuf::Map::swap() to swap two map fields.
890  * Fixed a memory leak when calling Reflection::ReleaseMessage() on a message
891    allocated on arena.
892  * Improved error reporting when parsing text format protos.
893  * JSON
894      - Added a new parser option to ignore unknown fields when parsing JSON.
895      - Added convenient methods for message to/from JSON conversion.
896  * Various performance optimizations.
897
898  Java (beta)
899  * File option "java_generate_equals_and_hash" is now deprecated. equals() and
900    hashCode() methods are generated by default.
901  * Added a new JSON printer option "omittingInsignificantWhitespace" to produce
902    a more compact JSON output. The printer will pretty-print by default.
903  * Updated Java runtime to be compatible with 2.5.0/2.6.1 generated protos.
904
905  Python (beta)
906  * Added support to pretty print Any messages in text format.
907  * Added a flag to ignore unknown fields when parsing JSON.
908  * Bugfix: "@type" field of a JSON Any message is now correctly put before
909    other fields.
910
911  Objective-C (beta)
912  * Updated the code to support compiling with more compiler warnings
913    enabled. (Issue 1616)
914  * Exposing more detailed errors for parsing failures. (PR 1623)
915  * Small (breaking) change to the naming of some methods on the support classes
916    for map<>. There were collisions with the system provided KVO support, so
917    the names were changed to avoid those issues.  (PR 1699)
918  * Fixed for proper Swift bridging of error handling during parsing. (PR 1712)
919  * Complete support for generating sources that will go into a Framework and
920    depend on generated sources from other Frameworks. (Issue 1457)
921
922  C# (beta)
923  * RepeatedField optimizations.
924  * Support for .NET Core.
925  * Minor bug fixes.
926  * Ability to format a single value in JsonFormatter (advanced usage only).
927  * Modifications to attributes applied to generated code.
928
929  Javascript (alpha)
930  * Maps now have a real map API instead of being treated as repeated fields.
931  * Well-known types are now provided in the google-protobuf package, and the
932    code generator knows to require() them from that package.
933  * Bugfix: non-canonical varints are correctly decoded.
934
935  Ruby (alpha)
936  * Accessors for oneof fields now return default values instead of nil.
937
938  Java Lite
939  * Java lite support is removed from protocol compiler. It will be supported
940    as a protocol compiler plugin in a separate code branch.
941
9422016-05-16 version 3.0.0-beta-3 (C++/Java/Python/Ruby/Nano/Objective-C/C#/JavaScript)
943  General
944  * Supported Proto3 lite-runtime in C++/Java for mobile platforms.
945  * Any type now supports APIs to specify prefixes other than
946    type.googleapis.com
947  * Removed javanano_use_deprecated_package option; Nano will always has its own
948    ".nano" package.
949
950  C++ (Beta)
951  * Improved hash maps.
952      - Improved hash maps comments. In particular, please note that equal hash
953        maps will not necessarily have the same iteration order and
954        serialization.
955      - Added a new hash maps implementation that will become the default in a
956        later release.
957  * Arenas
958      - Several inlined methods in Arena were moved to out-of-line to improve
959        build performance and code size.
960      - Added SpaceAllocatedAndUsed() to report both space used and allocated
961      - Added convenient class UnsafeArenaAllocatedRepeatedPtrFieldBackInserter
962  * Any
963      - Allow custom type URL prefixes in Any packing.
964      - TextFormat now expand the Any type rather than printing bytes.
965  * Performance optimizations and various bug fixes.
966
967  Java (Beta)
968  * Introduced an ExperimentalApi annotation. Annotated APIs are experimental
969    and are subject to change in a backward incompatible way in future releases.
970  * Introduced zero-copy serialization as an ExperimentalApi
971      - Introduction of the `ByteOutput` interface. This is similar to
972        `OutputStream` but provides semantics for lazy writing (i.e. no
973        immediate copy required) of fields that are considered to be immutable.
974      - `ByteString` now supports writing to a `ByteOutput`, which will directly
975        expose the internals of the `ByteString` (i.e. `byte[]` or `ByteBuffer`)
976        to the `ByteOutput` without copying.
977      - `CodedOutputStream` now supports writing to a `ByteOutput`. `ByteString`
978        instances that are too large to fit in the internal buffer will be
979        (lazily) written to the `ByteOutput` directly.
980      - This allows applications using large `ByteString` fields to avoid
981        duplication of these fields entirely. Such an application can supply a
982        `ByteOutput` that chains together the chunks received from
983        `CodedOutputStream` before forwarding them onto the IO system.
984  * Other related changes to `CodedOutputStream`
985      - Additional use of `sun.misc.Unsafe` where possible to perform fast
986        access to `byte[]` and `ByteBuffer` values and avoiding unnecessary
987        range checking.
988      - `ByteBuffer`-backed `CodedOutputStream` now writes directly to the
989        `ByteBuffer` rather than to an intermediate array.
990  * Improved lite-runtime.
991      - Lite protos now implement deep equals/hashCode/toString
992      - Significantly improved the performance of Builder#mergeFrom() and
993        Builder#mergeDelimitedFrom()
994  * Various bug fixes and small feature enhancement.
995      - Fixed stack overflow when in hashCode() for infinite recursive oneofs.
996      - Fixed the lazy field parsing in lite to merge rather than overwrite.
997      - TextFormat now supports reporting line/column numbers on errors.
998      - Updated to add appropriate @Override for better compiler errors.
999
1000  Python (Beta)
1001  * Added JSON format for Any, Struct, Value and ListValue
1002  * [ ] is now accepted for both repeated scalar fields and repeated message
1003    fields in text format parser.
1004  * Numerical field name is now supported in text format.
1005  * Added DiscardUnknownFields API for python protobuf message.
1006
1007  Objective-C (Beta)
1008  * Proto comments now come over as HeaderDoc comments in the generated sources
1009    so Xcode can pick them up and display them.
1010  * The library headers have been updated to use HeaderDoc comments so Xcode can
1011    pick them up and display them.
1012  * The per message and per field overhead in both generated code and runtime
1013    object sizes was reduced.
1014  * Generated code now include deprecated annotations when the proto file
1015    included them.
1016
1017  C# (Beta)
1018  In general: some changes are breaking, which require regenerating messages.
1019  Most user-written code will not be impacted *except* for the renaming of enum
1020  values.
1021
1022  * Allow custom type URL prefixes in `Any` packing, and ignore them when
1023    unpacking
1024  * `protoc` is now in a separate NuGet package (Google.Protobuf.Tools)
1025  * New option: `internal_access` to generate internal classes
1026  * Enum values are now PascalCased, and if there's a prefix which matches the
1027    name of the enum, that is removed (so an enum `COLOR` with a value
1028    `COLOR_BLUE` would generate a value of just `Blue`). An option
1029    (`legacy_enum_values`) is temporarily available to disable this, but the
1030    option will be removed for GA.
1031  * `json_name` option is now honored
1032  * If group tags are encountered when parsing, they are validated more
1033    thoroughly (although we don't support actual groups)
1034  * NuGet dependencies are better specified
1035  * Breaking: `Preconditions` is renamed to `ProtoPreconditions`
1036  * Breaking: `GeneratedCodeInfo` is renamed to `GeneratedClrTypeInfo`
1037  * `JsonFormatter` now allows writing to a `TextWriter`
1038  * New interface, `ICustomDiagnosticMessage` to allow more compact
1039    representations from `ToString`
1040  * `CodedInputStream` and `CodedOutputStream` now implement `IDisposable`,
1041    which simply disposes of the streams they were constructed with
1042  * Map fields no longer support null values (in line with other languages)
1043  * Improvements in JSON formatting and parsing
1044
1045  Javascript (Alpha)
1046  * Better support for "bytes" fields: bytes fields can be read as either a
1047    base64 string or UInt8Array (in environments where TypedArray is supported).
1048  * New support for CommonJS imports.  This should make it easier to use the
1049    JavaScript support in Node.js and tools like WebPack.  See js/README.md for
1050    more information.
1051  * Some significant internal refactoring to simplify and modularize the code.
1052
1053  Ruby (Alpha)
1054  * JSON serialization now properly uses camelCased names, with a runtime option
1055    that will preserve original names from .proto files instead.
1056  * Well-known types are now included in the distribution.
1057  * Release now includes binary gems for Windows, Mac, and Linux instead of just
1058    source gems.
1059  * Bugfix for serializing oneofs.
1060
1061  C++/Java Lite (Alpha)
1062    A new "lite" generator parameter was introduced in the protoc for C++ and
1063    Java for Proto3 syntax messages. Example usage:
1064
1065     ./protoc --cpp_out=lite:$OUTPUT_PATH foo.proto
1066
1067    The protoc will treat the current input and all the transitive dependencies
1068    as LITE. The same generator parameter must be used to generate the
1069    dependencies.
1070
1071    In Proto3 syntax files, "optimized_for=LITE_RUNTIME" is no longer supported.
1072
1073
10742015-12-30 version 3.0.0-beta-2 (C++/Java/Python/Ruby/Nano/Objective-C/C#/JavaScript)
1075  General
1076  * Introduced a new language implementation: JavaScript.
1077  * Added a new field option "json_name". By default proto field names are
1078    converted to "lowerCamelCase" in proto3 JSON format. This option can be
1079    used to override this behavior and specify a different JSON name for the
1080    field.
1081  * Added conformance tests to ensure implementations are following proto3 JSON
1082    specification.
1083
1084  C++ (Beta)
1085  * Various bug fixes and improvements to the JSON support utility:
1086      - Duplicate map keys in JSON are now rejected (i.e., translation will
1087        fail).
1088      - Fixed wire-format for google.protobuf.Value/ListValue.
1089      - Fixed precision loss when converting google.protobuf.Timestamp.
1090      - Fixed a bug when parsing invalid UTF-8 code points.
1091      - Fixed a memory leak.
1092      - Reduced call stack usage.
1093
1094  Java (Beta)
1095  * Cleaned up some unused methods on CodedOutputStream.
1096  * Presized lists for packed fields during parsing in the lite runtime to
1097    reduce allocations and improve performance.
1098  * Improved the performance of unknown fields in the lite runtime.
1099  * Introduced UnsafeByteStrings to support zero-copy ByteString creation.
1100  * Various bug fixes and improvements to the JSON support utility:
1101      - Fixed a thread-safety bug.
1102      - Added a new option “preservingProtoFieldNames” to JsonFormat.
1103      - Added a new option “includingDefaultValueFields” to JsonFormat.
1104      - Updated the JSON utility to comply with proto3 JSON specification.
1105
1106  Python (Beta)
1107  * Added proto3 JSON format utility. It includes support for all field types
1108    and a few well-known types except for Any and Struct.
1109  * Added runtime support for Any, Timestamp, Duration and FieldMask.
1110  * [ ] is now accepted for repeated scalar fields in text format parser.
1111  * Map fields now have proper O(1) performance for lookup/insert/delete
1112    when using the Python/C++ implementation. They were previously using O(n)
1113    search-based algorithms because the C++ reflection interface didn't
1114    support true map operations.
1115
1116  Objective-C (Beta)
1117  * Various bug-fixes and code tweaks to pass more strict compiler warnings.
1118  * Now has conformance test coverage and is passing all tests.
1119
1120  C# (Beta)
1121  * Various bug-fixes.
1122  * Code generation: Files generated in directories based on namespace.
1123  * Code generation: Include comments from .proto files in XML doc
1124    comments (naively)
1125  * Code generation: Change organization/naming of "reflection class" (access
1126    to file descriptor)
1127  * Code generation and library: Add Parser property to MessageDescriptor,
1128    and introduce a non-generic parser type.
1129  * Library: Added TypeRegistry to support JSON parsing/formatting of Any.
1130  * Library: Added Any.Pack/Unpack support.
1131  * Library: Implemented JSON parsing.
1132
1133  Javascript (Alpha)
1134  * Added proto3 support for JavaScript. The runtime is written in pure
1135    JavaScript and works in browsers and in Node.js. To generate JavaScript
1136    code for your proto, invoke protoc with "--js_out". See js/README.md
1137    for more build instructions.
1138
11392015-08-26 version 3.0.0-beta-1 (C++/Java/Python/Ruby/Nano/Objective-C/C#)
1140  About Beta
1141  * This is the first beta release of protobuf v3.0.0. Not all languages
1142    have reached beta stage. Languages not marked as beta are still in
1143    alpha (i.e., be prepared for API breaking changes).
1144
1145  General
1146  * Proto3 JSON is supported in several languages (fully supported in C++
1147    and Java, partially supported in Ruby/C#). The JSON spec is defined in
1148    the proto3 language guide:
1149
1150      https://developers.google.com/protocol-buffers/docs/proto3#json
1151
1152    We will publish a more detailed spec to define the exact behavior of
1153    proto3-conformant JSON serializers and parsers. Until then, do not rely
1154    on specific behaviors of the implementation if it’s not documented in
1155    the above spec. More specifically, the behavior is not yet finalized for
1156    the following:
1157      - Parsing invalid JSON input (e.g., input with trailing commas).
1158      - Non-camelCase names in JSON input.
1159      - The same field appears multiple times in JSON input.
1160      - JSON arrays contain “null” values.
1161      - The message has unknown fields.
1162
1163  * Proto3 now enforces strict UTF-8 checking. Parsing will fail if a string
1164    field contains non UTF-8 data.
1165
1166  C++ (Beta)
1167  * Introduced new utility functions/classes in the google/protobuf/util
1168    directory:
1169      - MessageDifferencer: compare two proto messages and report their
1170                            differences.
1171      - JsonUtil: support converting protobuf binary format to/from JSON.
1172      - TimeUtil: utility functions to work with well-known types Timestamp
1173                  and Duration.
1174      - FieldMaskUtil: utility functions to work with FieldMask.
1175
1176  * Performance optimization of arena construction and destruction.
1177  * Bug fixes for arena and maps support.
1178  * Changed to use cmake for Windows Visual Studio builds.
1179  * Added Bazel support.
1180
1181  Java (Beta)
1182  * Introduced a new util package that will be distributed as a separate
1183    artifact in maven. It contains:
1184      - JsonFormat: convert proto messages to/from JSON.
1185      - TimeUtil: utility functions to work with Timestamp and Duration.
1186      - FieldMaskUtil: utility functions to work with FieldMask.
1187
1188  * The static PARSER in each generated message is deprecated, and it will
1189    be removed in a future release. A static parser() getter is generated
1190    for each message type instead.
1191  * Performance optimizations for String fields serialization.
1192  * Performance optimizations for Lite runtime on Android:
1193      - Reduced allocations
1194      - Reduced method overhead after ProGuarding
1195      - Reduced code size after ProGuarding
1196
1197  Python (Alpha)
1198  * Removed legacy Python 2.5 support.
1199  * Moved to a single Python 2.x/3.x-compatible codebase, instead of using 2to3.
1200  * Fixed build/tests on Python 2.6, 2.7, 3.3, and 3.4.
1201      - Pure-Python works on all four.
1202      - Python/C++ implementation works on all but 3.4, due to changes in the
1203        Python/C++ API in 3.4.
1204  * Some preliminary work has been done to allow for multiple DescriptorPools
1205    with Python/C++.
1206
1207  Ruby (Alpha)
1208  * Many bugfixes:
1209      - fixed parsing/serialization of bytes, sint, sfixed types
1210      - other parser bugfixes
1211      - fixed memory leak affecting Ruby 2.2
1212
1213  JavaNano (Alpha)
1214  * JavaNano generated code now will be put in a nano package by default to
1215    avoid conflicts with Java generated code.
1216
1217  Objective-C (Alpha)
1218  * Added non-null markup to ObjC library. Requires SDK 8.4+ to build.
1219  * Many bugfixes:
1220      - Removed the class/enum filter.
1221      - Renamed some internal types to avoid conflicts with the well-known types
1222        protos.
1223      - Added missing support for parsing repeated primitive fields in packed or
1224        unpacked forms.
1225      - Added *Count for repeated and map<> fields to avoid auto-create when
1226        checking for them being set.
1227
1228  C# (Alpha)
1229  * Namespace changed to Google.Protobuf (and NuGet package will be named
1230    correspondingly).
1231  * Target platforms now .NET 4.5 and selected portable subsets only.
1232  * Removed lite runtime.
1233  * Reimplementation to use mutable message types.
1234  * Null references used to represent "no value" for message type fields.
1235  * Proto3 semantics supported; proto2 files are prohibited for C# codegen.
1236    Most proto3 features supported:
1237      - JSON formatting (a.k.a. serialization to JSON), including well-known
1238        types (except for Any).
1239      - Wrapper types mapped to nullable value types (or string/ByteString
1240        allowing nullability). JSON parsing is not supported yet.
1241      - maps
1242      - oneof
1243      - enum unknown value preservation
1244
12452015-05-25 version 3.0.0-alpha-3 (Objective-C/C#):
1246  General
1247  * Introduced two new language implementations (Objective-C, C#) to proto3.
1248  * Explicit "optional" keyword are disallowed in proto3 syntax, as fields are
1249    optional by default.
1250  * Group fields are no longer supported in proto3 syntax.
1251  * Changed repeated primitive fields to use packed serialization by default in
1252    proto3 (implemented for C++, Java, Python in this release).  The user can
1253    still disable packed serialization by setting packed to false for now.
1254  * Added well-known type protos (any.proto, empty.proto, timestamp.proto,
1255    duration.proto, etc.). Users can import and use these protos just like
1256    regular proto files. Additional runtime support will be added for them in
1257    future releases (in the form of utility helper functions, or having them
1258    replaced by language specific types in generated code).
1259  * Added a "reserved" keyword in both proto2 and proto3 syntax. User can use
1260    this keyword to declare reserved field numbers and names to prevent them
1261    from being reused by other fields in the same message.
1262
1263    To reserve field numbers, add a reserved declaration in your message:
1264
1265      message TestMessage {
1266        reserved 2, 15, 9 to 11, 3;
1267      }
1268
1269    This reserves field numbers 2, 3, 9, 10, 11 and 15. If a user uses any of
1270    these as field numbers, the protocol buffer compiler will report an error.
1271
1272    Field names can also be reserved:
1273
1274      message TestMessage {
1275        reserved "foo", "bar";
1276      }
1277
1278  * Various bug fixes since 3.0.0-alpha-2
1279
1280  Objective-C
1281    Objective-C includes a code generator and a native objective-c runtime
1282    library.  By adding “--objc_out” to protoc, the code generator will generate
1283    a header(*.pbobjc.h) and an implementation file(*.pbobjc.m) for each proto
1284    file.
1285
1286    In this first release, the generated interface provides: enums, messages,
1287    field support(single, repeated, map, oneof), proto2 and proto3 syntax
1288    support, parsing and serialization. It’s  compatible with ARC and non-ARC
1289    usage. Besides, user can also access it via the swift bridging header.
1290
1291    See objectivec/README.md for details.
1292
1293  C#
1294    * C# protobufs are based on project
1295      https://github.com/jskeet/protobuf-csharp-port. The original project was
1296      frozen and all the new development will happen here.
1297    * Codegen plugin for C# was completely rewritten to C++ and is now an
1298      integral part of protoc.
1299    * Some refactorings and cleanup has been applied to the C# runtime library.
1300    * Only proto2 is supported in C# at the moment, proto3 support is in
1301      progress and will likely bring significant breaking changes to the API.
1302
1303    See csharp/README.md for details.
1304
1305  C++
1306    * Added runtime support for Any type. To use Any in your proto file, first
1307      import the definition of Any:
1308
1309        // foo.proto
1310        import "google/protobuf/any.proto";
1311        message Foo {
1312          google.protobuf.Any any_field = 1;
1313        }
1314        message Bar {
1315          int32 value = 1;
1316        }
1317
1318      Then in C++ you can access the Any field using PackFrom()/UnpackTo()
1319      methods:
1320
1321        Foo foo;
1322        Bar bar = ...;
1323        foo.mutable_any_field()->PackFrom(bar);
1324        ...
1325        if (foo.any_field().IsType<Bar>()) {
1326          foo.any_field().UnpackTo(&bar);
1327          ...
1328        }
1329    * In text format, entries of a map field will be sorted by key.
1330
1331  Java
1332    * Continued optimizations on the lite runtime to improve performance for
1333      Android.
1334
1335  Python
1336    * Added map support.
1337      - maps now have a dict-like interface (msg.map_field[key] = value)
1338      - existing code that modifies maps via the repeated field interface
1339        will need to be updated.
1340
1341  Ruby
1342    * Improvements to RepeatedField's emulation of the Ruby Array API.
1343    * Various speedups and internal cleanups.
1344
13452015-02-26 version 3.0.0-alpha-2 (Python/Ruby/JavaNano):
1346  General
1347  * Introduced three new language implementations (Ruby, JavaNano, and
1348    Python) to proto3.
1349  * Various bug fixes since 3.0.0-alpha-1
1350
1351  Python:
1352    Python has received several updates, most notably support for proto3
1353    semantics in any .proto file that declares syntax="proto3".
1354    Messages declared in proto3 files no longer represent field presence
1355    for scalar fields (number, enums, booleans, or strings).  You can
1356    no longer call HasField() for such fields, and they are serialized
1357    based on whether they have a non-zero/empty/false value.
1358
1359    One other notable change is in the C++-accelerated implementation.
1360    Descriptor objects (which describe the protobuf schema and allow
1361    reflection over it) are no longer duplicated between the Python
1362    and C++ layers.  The Python descriptors are now simple wrappers
1363    around the C++ descriptors.  This change should significantly
1364    reduce the memory usage of programs that use a lot of message
1365    types.
1366
1367  Ruby:
1368    We have added proto3 support for Ruby via a native C extension.
1369
1370    The Ruby extension itself is included in the ruby/ directory, and details on
1371    building and installing the extension are in ruby/README.md. The extension
1372    will also be published as a Ruby gem. Code generator support is included as
1373    part of `protoc` with the `--ruby_out` flag.
1374
1375    The Ruby extension implements a user-friendly DSL to define message types
1376    (also generated by the code generator from `.proto` files).  Once a message
1377    type is defined, the user may create instances of the message that behave in
1378    ways idiomatic to Ruby. For example:
1379
1380    - Message fields are present as ordinary Ruby properties (getter method
1381      `foo` and setter method `foo=`).
1382    - Repeated field elements are stored in a container that acts like a native
1383      Ruby array, and map elements are stored in a container that acts like a
1384      native Ruby hashmap.
1385    - The usual well-known methods, such as `#to_s`, `#dup`, and the like, are
1386      present.
1387
1388    Unlike several existing third-party Ruby extensions for protobuf, this
1389    extension is built on a "strongly-typed" philosophy: message fields and
1390    array/map containers will throw exceptions eagerly when values of the
1391    incorrect type are inserted.
1392
1393    See ruby/README.md for details.
1394
1395  JavaNano:
1396    JavaNano is a special code generator and runtime library designed especially
1397    for resource-restricted systems, like Android. It is very resource-friendly
1398    in both the amount of code and the runtime overhead. Here is an an overview
1399    of JavaNano features compared with the official Java protobuf:
1400
1401    - No descriptors or message builders.
1402    - All messages are mutable; fields are public Java fields.
1403    - For optional fields only, encapsulation behind setter/getter/hazzer/
1404      clearer functions is opt-in, which provide proper 'has' state support.
1405    - For proto2, if not opted in, has state (field presence) is not available.
1406      Serialization outputs all fields not equal to their defaults.
1407      The behavior is consistent with proto3 semantics.
1408    - Required fields (proto2 only) are always serialized.
1409    - Enum constants are integers; protection against invalid values only
1410      when parsing from the wire.
1411    - Enum constants can be generated into container interfaces bearing
1412      the enum's name (so the referencing code is in Java style).
1413    - CodedInputByteBufferNano can only take byte[] (not InputStream).
1414    - Similarly CodedOutputByteBufferNano can only write to byte[].
1415    - Repeated fields are in arrays, not ArrayList or Vector. Null array
1416      elements are allowed and silently ignored.
1417    - Full support for serializing/deserializing repeated packed fields.
1418    - Support  extensions (in proto2).
1419    - Unset messages/groups are null, not an immutable empty default
1420      instance.
1421    - toByteArray(...) and mergeFrom(...) are now static functions of
1422      MessageNano.
1423    - The 'bytes' type translates to the Java type byte[].
1424
1425    See javanano/README.txt for details.
1426
14272014-12-01 version 3.0.0-alpha-1 (C++/Java):
1428
1429  General
1430  * Introduced Protocol Buffers language version 3 (aka proto3).
1431
1432    When protobuf was initially opensourced it implemented Protocol Buffers
1433    language version 2 (aka proto2), which is why the version number
1434    started from v2.0.0. From v3.0.0, a new language version (proto3) is
1435    introduced while the old version (proto2) will continue to be supported.
1436
1437    The main intent of introducing proto3 is to clean up protobuf before
1438    pushing the language as the foundation of Google's new API platform.
1439    In proto3, the language is simplified, both for ease of use and  to
1440    make it available in a wider range of programming languages. At the
1441    same time a few features are added to better support common idioms
1442    found in APIs.
1443
1444    The following are the main new features in language version 3:
1445
1446      1. Removal of field presence logic for primitive value fields, removal
1447         of required fields, and removal of default values. This makes proto3
1448         significantly easier to implement with open struct representations,
1449         as in languages like Android Java, Objective C, or Go.
1450      2. Removal of unknown fields.
1451      3. Removal of extensions, which are instead replaced by a new standard
1452         type called Any.
1453      4. Fix semantics for unknown enum values.
1454      5. Addition of maps.
1455      6. Addition of a small set of standard types for representation of time,
1456         dynamic data, etc.
1457      7. A well-defined encoding in JSON as an alternative to binary proto
1458         encoding.
1459
1460    This release (v3.0.0-alpha-1) includes partial proto3 support for C++ and
1461    Java. Items 6 (well-known types) and 7 (JSON format) in the above feature
1462    list are not implemented.
1463
1464    A new notion "syntax" is introduced to specify whether a .proto file
1465    uses proto2 or proto3:
1466
1467      // foo.proto
1468      syntax = "proto3";
1469      message Bar {...}
1470
1471    If omitted, the protocol compiler will generate a warning and "proto2" will
1472    be used as the default. This warning will be turned into an error in a
1473    future release.
1474
1475    We recommend that new Protocol Buffers users use proto3. However, we do not
1476    generally recommend that existing users migrate from proto2 from proto3 due
1477    to API incompatibility, and we will continue to support proto2 for a long
1478    time.
1479
1480  * Added support for map fields (implemented in C++/Java for both proto2 and
1481    proto3).
1482
1483    Map fields can be declared using the following syntax:
1484
1485      message Foo {
1486        map<string, string> values = 1;
1487      }
1488
1489    Data of a map field will be stored in memory as an unordered map and it
1490    can be accessed through generated accessors.
1491
1492  C++
1493  * Added arena allocation support (for both proto2 and proto3).
1494
1495    Profiling shows memory allocation and deallocation constitutes a significant
1496    fraction of CPU-time spent in protobuf code and arena allocation is a
1497    technique introduced to reduce this cost. With arena allocation, new
1498    objects will be allocated from a large piece of preallocated memory and
1499    deallocation of these objects is almost free. Early adoption shows 20% to
1500    50% improvement in some Google binaries.
1501
1502    To enable arena support, add the following option to your .proto file:
1503
1504      option cc_enable_arenas = true;
1505
1506    Protocol compiler will generate additional code to make the generated
1507    message classes work with arenas. This does not change the existing API
1508    of protobuf messages and does not affect wire format. Your existing code
1509    should continue to work after adding this option. In the future we will
1510    make this option enabled by default.
1511
1512    To actually take advantage of arena allocation, you need to use the arena
1513    APIs when creating messages. A quick example of using the arena API:
1514
1515      {
1516        google::protobuf::Arena arena;
1517        // Allocate a protobuf message in the arena.
1518        MyMessage* message = Arena::CreateMessage<MyMessage>(&arena);
1519        // All submessages will be allocated in the same arena.
1520        if (!message->ParseFromString(data)) {
1521          // Deal with malformed input data.
1522        }
1523        // Must not delete the message here. It will be deleted automatically
1524        // when the arena is destroyed.
1525      }
1526
1527    Currently arena does not work with map fields. Enabling arena in a .proto
1528    file containing map fields will result in compile errors in the generated
1529    code. This will be addressed in a future release.
1530
15312014-10-20 version 2.6.1:
1532
1533  C++
1534  * Added atomicops support for Solaris.
1535  * Released memory allocated by InitializeDefaultRepeatedFields() and
1536    GetEmptyString(). Some memory sanitizers reported them as memory leaks.
1537
1538  Java
1539  * Updated DynamicMessage.setField() to handle repeated enum values
1540    correctly.
1541  * Fixed a bug that caused NullPointerException to be thrown when
1542    converting manually constructed FileDescriptorProto to
1543    FileDescriptor.
1544
1545  Python
1546  * Fixed WhichOneof() to work with de-serialized protobuf messages.
1547  * Fixed a missing file problem of Python C++ implementation.
1548
15492014-08-15 version 2.6.0:
1550
1551  General
1552  * Added oneofs(unions) feature. Fields in the same oneof will share
1553    memory and at most one field can be set at the same time. Use the
1554    oneof keyword to define a oneof like:
1555      message SampleMessage {
1556        oneof test_oneof {
1557          string name = 4;
1558          YourMessage sub_message = 9;
1559        }
1560      }
1561  * Files, services, enums, messages, methods and enum values can be marked
1562    as deprecated now.
1563  * Added Support for list values, including lists of messages, when
1564    parsing text-formatted protos in C++ and Java.
1565      For example:  foo: [1, 2, 3]
1566
1567  C++
1568  * Enhanced customization on TestFormat printing.
1569  * Added SwapFields() in reflection API to swap a subset of fields.
1570    Added SetAllocatedMessage() in reflection API.
1571  * Repeated primitive extensions are now packable. The
1572    [packed=true] option only affects serializers. Therefore, it is
1573    possible to switch a repeated extension field to packed format
1574    without breaking backwards-compatibility.
1575  * Various speed optimizations.
1576
1577  Java
1578  * writeTo() method in ByteString can now write a substring to an
1579    output stream. Added endWith() method for ByteString.
1580  * ByteString and ByteBuffer are now supported in CodedInputStream
1581    and CodedOutputStream.
1582  * java_generate_equals_and_hash can now be used with the LITE_RUNTIME.
1583
1584  Python
1585  * A new C++-backed extension module (aka "cpp api v2") that replaces the
1586    old ("cpp api v1") one.  Much faster than the pure Python code.  This one
1587    resolves many bugs and is recommended for general use over the
1588    pure Python when possible.
1589  * Descriptors now have enum_types_by_name and extension_types_by_name dict
1590    attributes.
1591  * Support for Python 3.
1592
15932013-02-27 version 2.5.0:
1594
1595  General
1596  * New notion "import public" that allows a proto file to forward the content
1597    it imports to its importers. For example,
1598      // foo.proto
1599      import public "bar.proto";
1600      import "baz.proto";
1601
1602      // qux.proto
1603      import "foo.proto";
1604      // Stuff defined in bar.proto may be used in this file, but stuff from
1605      // baz.proto may NOT be used without importing it explicitly.
1606    This is useful for moving proto files. To move a proto file, just leave
1607    a single "import public" in the old proto file.
1608  * New enum option "allow_alias" that specifies whether different symbols can
1609    be assigned the same numeric value. Default value is "true". Setting it to
1610    false causes the compiler to reject enum definitions where multiple symbols
1611    have the same numeric value.
1612    Note: We plan to flip the default value to "false" in a future release.
1613    Projects using enum aliases should set the option to "true" in their .proto
1614    files.
1615
1616  C++
1617  * New generated method set_allocated_foo(Type* foo) for message and string
1618    fields. This method allows you to set the field to a pre-allocated object
1619    and the containing message takes the ownership of that object.
1620  * Added SetAllocatedExtension() and ReleaseExtension() to extensions API.
1621  * Custom options are now formatted correctly when descriptors are printed in
1622    text format.
1623  * Various speed optimizations.
1624
1625  Java
1626  * Comments in proto files are now collected and put into generated code as
1627    comments for corresponding classes and data members.
1628  * Added Parser to parse directly into messages without a Builder. For
1629    example,
1630      Foo foo = Foo.PARSER.ParseFrom(input);
1631    Using Parser is ~25% faster than using Builder to parse messages.
1632  * Added getters/setters to access the underlying ByteString of a string field
1633    directly.
1634  * ByteString now supports more operations: substring(), prepend(), and
1635    append(). The implementation of ByteString uses a binary tree structure
1636    to support these operations efficiently.
1637  * New method findInitializationErrors() that lists all missing required
1638    fields.
1639  * Various code size and speed optimizations.
1640
1641  Python
1642  * Added support for dynamic message creation. DescriptorDatabase,
1643    DescriptorPool, and MessageFactory work like their C++ counterparts to
1644    simplify Descriptor construction from *DescriptorProtos, and MessageFactory
1645    provides a message instance from a Descriptor.
1646  * Added pickle support for protobuf messages.
1647  * Unknown fields are now preserved after parsing.
1648  * Fixed bug where custom options were not correctly populated. Custom
1649    options can be accessed now.
1650  * Added EnumTypeWrapper that provides better accessibility to enum types.
1651  * Added ParseMessage(descriptor, bytes) to generate a new Message instance
1652    from a descriptor and a byte string.
1653
16542011-05-01 version 2.4.1:
1655
1656  C++
1657  * Fixed the friendship problem for old compilers to make the library now gcc 3
1658    compatible again.
1659  * Fixed vcprojects/extract_includes.bat to extract compiler/plugin.h.
1660
1661  Java
1662  * Removed usages of JDK 1.6 only features to make the library now JDK 1.5
1663    compatible again.
1664  * Fixed a bug about negative enum values.
1665  * serialVersionUID is now defined in generated messages for java serializing.
1666  * Fixed protoc to use java.lang.Object, which makes "Object" now a valid
1667    message name again.
1668
1669  Python
1670  * Experimental C++ implementation now requires C++ protobuf library installed.
1671    See the README.txt in the python directory for details.
1672
16732011-02-02 version 2.4.0:
1674
1675  General
1676  * The RPC (cc|java|py)_generic_services default value is now false instead of
1677    true.
1678  * Custom options can have aggregate types. For example,
1679      message MyOption {
1680        optional string comment = 1;
1681        optional string author = 2;
1682      }
1683      extend google.protobuf.FieldOptions {
1684        optional MyOption myoption = 12345;
1685      }
1686    This option can now be set as follows:
1687      message SomeType {
1688        optional int32 field = 1 [(myoption) = { comment:'x' author:'y' }];
1689      }
1690
1691  C++
1692  * Various speed and code size optimizations.
1693  * Added a release_foo() method on string and message fields.
1694  * Fixed gzip_output_stream sub-stream handling.
1695
1696  Java
1697  * Builders now maintain sub-builders for sub-messages. Use getFooBuilder() to
1698    get the builder for the sub-message "foo". This allows you to repeatedly
1699    modify deeply-nested sub-messages without rebuilding them.
1700  * Builder.build() no longer invalidates the Builder for generated messages
1701    (You may continue to modify it and then build another message).
1702  * Code generator will generate efficient equals() and hashCode()
1703    implementations if new option java_generate_equals_and_hash is enabled.
1704    (Otherwise, reflection-based implementations are used.)
1705  * Generated messages now implement Serializable.
1706  * Fields with [deprecated=true] will be marked with @Deprecated in Java.
1707  * Added lazy conversion of UTF-8 encoded strings to String objects to improve
1708    performance.
1709  * Various optimizations.
1710  * Enum value can be accessed directly, instead of calling getNumber() on the
1711    enum member.
1712  * For each enum value, an integer constant is also generated with the suffix
1713    _VALUE.
1714
1715  Python
1716  * Added an experimental  C++ implementation for Python messages via a Python
1717    extension. Implementation type is controlled by an environment variable
1718    PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION (valid values: "cpp" and "python")
1719    The default value is currently "python" but will be changed to "cpp" in
1720    future release.
1721  * Improved performance on message instantiation significantly.
1722    Most of the work on message instantiation is done just once per message
1723    class, instead of once per message instance.
1724  * Improved performance on text message parsing.
1725  * Allow add() to forward keyword arguments to the concrete class.
1726      E.g. instead of
1727        item = repeated_field.add()
1728        item.foo = bar
1729        item.baz = quux
1730      You can do:
1731        repeated_field.add(foo=bar, baz=quux)
1732  * Added a sort() interface to the BaseContainer.
1733  * Added an extend() method to repeated composite fields.
1734  * Added UTF8 debug string support.
1735
17362010-01-08 version 2.3.0:
1737
1738  General
1739  * Parsers for repeated numeric fields now always accept both packed and
1740    unpacked input.  The [packed=true] option only affects serializers.
1741    Therefore, it is possible to switch a field to packed format without
1742    breaking backwards-compatibility -- as long as all parties are using
1743    protobuf 2.3.0 or above, at least.
1744  * The generic RPC service code generated by the C++, Java, and Python
1745    generators can be disabled via file options:
1746      option cc_generic_services = false;
1747      option java_generic_services = false;
1748      option py_generic_services = false;
1749    This allows plugins to generate alternative code, possibly specific to some
1750    particular RPC implementation.
1751
1752  protoc
1753  * Now supports a plugin system for code generators.  Plugins can generate
1754    code for new languages or inject additional code into the output of other
1755    code generators.  Plugins are just binaries which accept a protocol buffer
1756    on stdin and write a protocol buffer to stdout, so they may be written in
1757    any language.  See src/google/protobuf/compiler/plugin.proto.
1758    **WARNING**:  Plugins are experimental.  The interface may change in a
1759    future version.
1760  * If the output location ends in .zip or .jar, protoc will write its output
1761    to a zip/jar archive instead of a directory.  For example:
1762      protoc --java_out=myproto_srcs.jar --python_out=myproto.zip myproto.proto
1763    Currently the archive contents are not compressed, though this could change
1764    in the future.
1765  * inf, -inf, and nan can now be used as default values for float and double
1766    fields.
1767
1768  C++
1769  * Various speed and code size optimizations.
1770  * DynamicMessageFactory is now fully thread-safe.
1771  * Message::Utf8DebugString() method is like DebugString() but avoids escaping
1772    UTF-8 bytes.
1773  * Compiled-in message types can now contain dynamic extensions, through use
1774    of CodedInputStream::SetExtensionRegistry().
1775  * Now compiles shared libraries (DLLs) by default on Cygwin and MinGW, to
1776    match other platforms.  Use --disable-shared to avoid this.
1777
1778  Java
1779  * parseDelimitedFrom() and mergeDelimitedFrom() now detect EOF and return
1780    false/null instead of throwing an exception.
1781  * Fixed some initialization ordering bugs.
1782  * Fixes for OpenJDK 7.
1783
1784  Python
1785  * 10-25 times faster than 2.2.0, still pure-Python.
1786  * Calling a mutating method on a sub-message always instantiates the message
1787    in its parent even if the mutating method doesn't actually mutate anything
1788    (e.g. parsing from an empty string).
1789  * Expanded descriptors a bit.
1790
17912009-08-11 version 2.2.0:
1792
1793  C++
1794  * Lite mode:  The "optimize_for = LITE_RUNTIME" option causes the compiler
1795    to generate code which only depends libprotobuf-lite, which is much smaller
1796    than libprotobuf but lacks descriptors, reflection, and some other features.
1797  * Fixed bug where Message.Swap(Message) was only implemented for
1798    optimize_for_speed.  Swap now properly implemented in both modes
1799    (Issue 91).
1800  * Added RemoveLast and SwapElements(index1, index2) to Reflection
1801    interface for repeated elements.
1802  * Added Swap(Message) to Reflection interface.
1803  * Floating-point literals in generated code that are intended to be
1804    single-precision now explicitly have 'f' suffix to avoid pedantic warnings
1805    produced by some compilers.
1806  * The [deprecated=true] option now causes the C++ code generator to generate
1807    a GCC-style deprecation annotation (no-op on other compilers).
1808  * google::protobuf::GetEnumDescriptor<SomeGeneratedEnumType>() returns the
1809    EnumDescriptor for that type -- useful for templates which cannot call
1810    SomeGeneratedEnumType_descriptor().
1811  * Various optimizations and obscure bug fixes.
1812
1813  Java
1814  * Lite mode:  The "optimize_for = LITE_RUNTIME" option causes the compiler
1815    to generate code which only depends libprotobuf-lite, which is much smaller
1816    than libprotobuf but lacks descriptors, reflection, and some other features.
1817  * Lots of style cleanups.
1818
1819  Python
1820  * Fixed endianness bug with floats and doubles.
1821  * Text format parsing support.
1822  * Fix bug with parsing packed repeated fields in embedded messages.
1823  * Ability to initialize fields by passing keyword args to constructor.
1824  * Support iterators in extend and __setslice__ for containers.
1825
18262009-05-13 version 2.1.0:
1827
1828  General
1829  * Repeated fields of primitive types (types other that string, group, and
1830    nested messages) may now use the option [packed = true] to get a more
1831    efficient encoding.  In the new encoding, the entire list is written
1832    as a single byte blob using the "length-delimited" wire type.  Within
1833    this blob, the individual values are encoded the same way they would
1834    be normally except without a tag before each value (thus, they are
1835    tightly "packed").
1836  * For each field, the generated code contains an integer constant assigned
1837    to the field number.  For example, the .proto file:
1838      message Foo { optional int bar_baz = 123; }
1839    would generate the following constants, all with the integer value 123:
1840      C++:     Foo::kBarBazFieldNumber
1841      Java:    Foo.BAR_BAZ_FIELD_NUMBER
1842      Python:  Foo.BAR_BAZ_FIELD_NUMBER
1843    Constants are also generated for extensions, with the same naming scheme.
1844    These constants may be used as switch cases.
1845  * Updated bundled Google Test to version 1.3.0.  Google Test is now bundled
1846    in its verbatim form as a nested autoconf package, so you can drop in any
1847    other version of Google Test if needed.
1848  * optimize_for = SPEED is now the default, by popular demand.  Use
1849    optimize_for = CODE_SIZE if code size is more important in your app.
1850  * It is now an error to define a default value for a repeated field.
1851    Previously, this was silently ignored (it had no effect on the generated
1852    code).
1853  * Fields can now be marked deprecated like:
1854      optional int32 foo = 1 [deprecated = true];
1855    Currently this does not have any actual effect, but in the future the code
1856    generators may generate deprecation annotations in each language.
1857  * Cross-compiling should now be possible using the --with-protoc option to
1858    configure.  See README.txt for more info.
1859
1860  protoc
1861  * --error_format=msvs option causes errors to be printed in Visual Studio
1862    format, which should allow them to be clicked on in the build log to go
1863    directly to the error location.
1864  * The type name resolver will no longer resolve type names to fields.  For
1865    example, this now works:
1866      message Foo {}
1867      message Bar {
1868        optional int32 Foo = 1;
1869        optional Foo baz = 2;
1870      }
1871    Previously, the type of "baz" would resolve to "Bar.Foo", and you'd get
1872    an error because Bar.Foo is a field, not a type.  Now the type of "baz"
1873    resolves to the message type Foo.  This change is unlikely to make a
1874    difference to anyone who follows the Protocol Buffers style guide.
1875
1876  C++
1877  * Several optimizations, including but not limited to:
1878    - Serialization, especially to flat arrays, is 10%-50% faster, possibly
1879      more for small objects.
1880    - Several descriptor operations which previously required locking no longer
1881      do.
1882    - Descriptors are now constructed lazily on first use, rather than at
1883      process startup time.  This should save memory in programs which do not
1884      use descriptors or reflection.
1885    - UnknownFieldSet completely redesigned to be more efficient (especially in
1886      terms of memory usage).
1887    - Various optimizations to reduce code size (though the serialization speed
1888      optimizations increased code size).
1889  * Message interface has method ParseFromBoundedZeroCopyStream() which parses
1890    a limited number of bytes from an input stream rather than parsing until
1891    EOF.
1892  * GzipInputStream and GzipOutputStream support reading/writing gzip- or
1893    zlib-compressed streams if zlib is available.
1894    (google/protobuf/io/gzip_stream.h)
1895  * DescriptorPool::FindAllExtensions() and corresponding
1896    DescriptorDatabase::FindAllExtensions() can be used to enumerate all
1897    extensions of a given type.
1898  * For each enum type Foo, protoc will generate functions:
1899      const string& Foo_Name(Foo value);
1900      bool Foo_Parse(const string& name, Foo* result);
1901    The former returns the name of the enum constant corresponding to the given
1902    value while the latter finds the value corresponding to a name.
1903  * RepeatedField and RepeatedPtrField now have back-insertion iterators.
1904  * String fields now have setters that take a char* and a size, in addition
1905    to the existing ones that took char* or const string&.
1906  * DescriptorPool::AllowUnknownDependencies() may be used to tell
1907    DescriptorPool to create placeholder descriptors for unknown entities
1908    referenced in a FileDescriptorProto.  This can allow you to parse a .proto
1909    file without having access to other .proto files that it imports, for
1910    example.
1911  * Updated gtest to latest version.  The gtest package is now included as a
1912    nested autoconf package, so it should be able to drop new versions into the
1913    "gtest" subdirectory without modification.
1914
1915  Java
1916  * Fixed bug where Message.mergeFrom(Message) failed to merge extensions.
1917  * Message interface has new method toBuilder() which is equivalent to
1918    newBuilderForType().mergeFrom(this).
1919  * All enums now implement the ProtocolMessageEnum interface.
1920  * Setting a field to null now throws NullPointerException.
1921  * Fixed tendency for TextFormat's parsing to overflow the stack when
1922    parsing large string values.  The underlying problem is with Java's
1923    regex implementation (which unfortunately uses recursive backtracking
1924    rather than building an NFA).  Worked around by making use of possessive
1925    quantifiers.
1926  * Generated service classes now also generate pure interfaces.  For a service
1927    Foo, Foo.Interface is a pure interface containing all of the service's
1928    defined methods.  Foo.newReflectiveService() can be called to wrap an
1929    instance of this interface in a class that implements the generic
1930    RpcService interface, which provides reflection support that is usually
1931    needed by RPC server implementations.
1932  * RPC interfaces now support blocking operation in addition to non-blocking.
1933    The protocol compiler generates separate blocking and non-blocking stubs
1934    which operate against separate blocking and non-blocking RPC interfaces.
1935    RPC implementations will have to implement the new interfaces in order to
1936    support blocking mode.
1937  * New I/O methods parseDelimitedFrom(), mergeDelimitedFrom(), and
1938    writeDelimitedTo() read and write "delimited" messages from/to a stream,
1939    meaning that the message size precedes the data.  This way, you can write
1940    multiple messages to a stream without having to worry about delimiting
1941    them yourself.
1942  * Throw a more descriptive exception when build() is double-called.
1943  * Add a method to query whether CodedInputStream is at the end of the input
1944    stream.
1945  * Add a method to reset a CodedInputStream's size counter; useful when
1946    reading many messages with the same stream.
1947  * equals() and hashCode() now account for unknown fields.
1948
1949  Python
1950  * Added slicing support for repeated scalar fields. Added slice retrieval and
1951    removal of repeated composite fields.
1952  * Updated RPC interfaces to allow for blocking operation.  A client may
1953    now pass None for a callback when making an RPC, in which case the
1954    call will block until the response is received, and the response
1955    object will be returned directly to the caller.  This interface change
1956    cannot be used in practice until RPC implementations are updated to
1957    implement it.
1958  * Changes to input_stream.py should make protobuf compatible with appengine.
1959
19602008-11-25 version 2.0.3:
1961
1962  protoc
1963  * Enum values may now have custom options, using syntax similar to field
1964    options.
1965  * Fixed bug where .proto files which use custom options but don't actually
1966    define them (i.e. they import another .proto file defining the options)
1967    had to explicitly import descriptor.proto.
1968  * Adjacent string literals in .proto files will now be concatenated, like in
1969    C.
1970  * If an input file is a Windows absolute path (e.g. "C:\foo\bar.proto") and
1971    the import path only contains "." (or contains "." but does not contain
1972    the file), protoc incorrectly thought that the file was under ".", because
1973    it thought that the path was relative (since it didn't start with a slash).
1974    This has been fixed.
1975
1976  C++
1977  * Generated message classes now have a Swap() method which efficiently swaps
1978    the contents of two objects.
1979  * All message classes now have a SpaceUsed() method which returns an estimate
1980    of the number of bytes of allocated memory currently owned by the object.
1981    This is particularly useful when you are reusing a single message object
1982    to improve performance but want to make sure it doesn't bloat up too large.
1983  * New method Message::SerializeAsString() returns a string containing the
1984    serialized data.  May be more convenient than calling
1985    SerializeToString(string*).
1986  * In debug mode, log error messages when string-type fields are found to
1987    contain bytes that are not valid UTF-8.
1988  * Fixed bug where a message with multiple extension ranges couldn't parse
1989    extensions.
1990  * Fixed bug where MergeFrom(const Message&) didn't do anything if invoked on
1991    a message that contained no fields (but possibly contained extensions).
1992  * Fixed ShortDebugString() to not be O(n^2).  Durr.
1993  * Fixed crash in TextFormat parsing if the first token in the input caused a
1994    tokenization error.
1995  * Fixed obscure bugs in zero_copy_stream_impl.cc.
1996  * Added support for HP C++ on Tru64.
1997  * Only build tests on "make check", not "make".
1998  * Fixed alignment issue that caused crashes when using DynamicMessage on
1999    64-bit Sparc machines.
2000  * Simplify template usage to work with MSVC 2003.
2001  * Work around GCC 4.3.x x86_64 compiler bug that caused crashes on startup.
2002    (This affected Fedora 9 in particular.)
2003  * Now works on "Solaris 10 using recent Sun Studio".
2004
2005  Java
2006  * New overload of mergeFrom() which parses a slice of a byte array instead
2007    of the whole thing.
2008  * New method ByteString.asReadOnlyByteBuffer() does what it sounds like.
2009  * Improved performance of isInitialized() when optimizing for code size.
2010
2011  Python
2012  * Corrected ListFields() signature in Message base class to match what
2013    subclasses actually implement.
2014  * Some minor refactoring.
2015  * Don't pass self as first argument to superclass constructor (no longer
2016    allowed in Python 2.6).
2017
20182008-09-29 version 2.0.2:
2019
2020  General
2021  * License changed from Apache 2.0 to 3-Clause BSD.
2022  * It is now possible to define custom "options", which are basically
2023    annotations which may be placed on definitions in a .proto file.
2024    For example, you might define a field option called "foo" like so:
2025      import "google/protobuf/descriptor.proto"
2026      extend google.protobuf.FieldOptions {
2027        optional string foo = 12345;
2028      }
2029    Then you annotate a field using the "foo" option:
2030      message MyMessage {
2031        optional int32 some_field = 1 [(foo) = "bar"]
2032      }
2033    The value of this option is then visible via the message's
2034    Descriptor:
2035      const FieldDescriptor* field =
2036        MyMessage::descriptor()->FindFieldByName("some_field");
2037      assert(field->options().GetExtension(foo) == "bar");
2038    This feature has been implemented and tested in C++ and Java.
2039    Other languages may or may not need to do extra work to support
2040    custom options, depending on how they construct descriptors.
2041
2042  C++
2043  * Fixed some GCC warnings that only occur when using -pedantic.
2044  * Improved static initialization code, making ordering more
2045    predictable among other things.
2046  * TextFormat will no longer accept messages which contain multiple
2047    instances of a singular field.  Previously, the latter instance
2048    would overwrite the former.
2049  * Now works on systems that don't have hash_map.
2050
2051  Java
2052  * Print @Override annotation in generated code where appropriate.
2053
2054  Python
2055  * Strings now use the "unicode" type rather than the "str" type.
2056    String fields may still be assigned ASCII "str" values; they will
2057    automatically be converted.
2058  * Adding a property to an object representing a repeated field now
2059    raises an exception.  For example:
2060      # No longer works (and never should have).
2061      message.some_repeated_field.foo = 1
2062
2063  Windows
2064  * We now build static libraries rather than DLLs by default on MSVC.
2065    See vsprojects/readme.txt for more information.
2066
20672008-08-15 version 2.0.1:
2068
2069  protoc
2070  * New flags --encode and --decode can be used to convert between protobuf text
2071    format and binary format from the command-line.
2072  * New flag --descriptor_set_out can be used to write FileDescriptorProtos for
2073    all parsed files directly into a single output file.  This is particularly
2074    useful if you wish to parse .proto files from programs written in languages
2075    other than C++: just run protoc as a background process and have it output
2076    a FileDescriptorList, then parse that natively.
2077  * Improved error message when an enum value's name conflicts with another
2078    symbol defined in the enum type's scope, e.g. if two enum types declared
2079    in the same scope have values with the same name.  This is disallowed for
2080    compatibility with C++, but this wasn't clear from the error.
2081  * Fixed absolute output paths on Windows.
2082  * Allow trailing slashes in --proto_path mappings.
2083
2084  C++
2085  * Reflection objects are now per-class rather than per-instance.  To make this
2086    possible, the Reflection interface had to be changed such that all methods
2087    take the Message instance as a parameter.  This change improves performance
2088    significantly in memory-bandwidth-limited use cases, since it makes the
2089    message objects smaller.  Note that source-incompatible interface changes
2090    like this will not be made again after the library leaves beta.
2091  * Heuristically detect sub-messages when printing unknown fields.
2092  * Fix static initialization ordering bug that caused crashes at startup when
2093    compiling on Mac with static linking.
2094  * Fixed TokenizerTest when compiling with -DNDEBUG on Linux.
2095  * Fixed incorrect definition of kint32min.
2096  * Fix bytes type setter to work with byte sequences with embedded NULLs.
2097  * Other irrelevant tweaks.
2098
2099  Java
2100  * Fixed UnknownFieldSet's parsing of varints larger than 32 bits.
2101  * Fixed TextFormat's parsing of "inf" and "nan".
2102  * Fixed TextFormat's parsing of comments.
2103  * Added info to Java POM that will be required when we upload the
2104    package to a Maven repo.
2105
2106  Python
2107  * MergeFrom(message) and CopyFrom(message) are now implemented.
2108  * SerializeToString() raises an exception if the message is missing required
2109    fields.
2110  * Code organization improvements.
2111  * Fixed doc comments for RpcController and RpcChannel, which had somehow been
2112    swapped.
2113  * Fixed text_format_test on Windows where floating-point exponents sometimes
2114    contain extra zeros.
2115  * Fix Python service CallMethod() implementation.
2116
2117  Other
2118  * Improved readmes.
2119  * VIM syntax highlighting improvements.
2120
21212008-07-07 version 2.0.0:
2122
2123  * First public release.
2124