Skip to content

Releases: fxamacker/cbor

v2.7.0

24 Jun 04:56
02b69db
Compare
Choose a tag to compare

v2.7.0 (June 23, 2024)

v2.7.0 adds features and improvements that help large projects (e.g. Kubernetes) use CBOR as an alternative to JSON and Protocol Buffers. Other improvements include speedups, improved memory use, bug fixes, decoding/encoding options, etc. Passed 5+ billion execs fuzzing.

Special thanks to @benluddy for contributing features, optimizations, improvements, bug fixes, and discussions! 👍

⭐ Features and Optimizations

  • Add decoding option TimeTagToAny to produce RFC3339 when decoding time into interface{} by @ssuriyan7 in #506
  • Bypass sorting overhead for single-entry maps. by @benluddy in #516
  • Add decode option to allow rejecting inputs that contain certain simple values. by @benluddy in #481
  • Allow rejection of NaN and Inf float values on encode and decode. by @benluddy in #513
  • Add SortMode to encode struct fields in a less predictable order. by @benluddy in #515
  • Add a decoding option to allow decoding byte string into time.Time. by @benluddy in #524
  • Encode structs directly to output buffer. by @benluddy in #519
  • Support automatic conversion between text and binary string representations by @benluddy in #476
  • Add a method for marshaling directly into a user-provided buffer. by @benluddy in #521
  • Add options to disable BinaryMarshaler/BinaryUnmarshaler support. by @benluddy in #526
  • Add option to reject decoding bignum tags and encoding big.Int. by @benluddy in #527
  • Improve speed & memory use for Diagnose() and DiagnoseFirst() by @benluddy in #533
  • Refactor sorted map encode to use fewer buffers for nested maps. by @benluddy in #537
  • Allow user to specify buffer by adding cbor.MarshalToBuffer(), UserBufferEncMode interface, etc. by @fxamacker in #553

🛠 Improvements

Most improvements focused on fixing or updating new features. See 🐞 Bug Fixes section for fixes to features already released.

  • Refactor to reuse functions and improve code coverage by @fxamacker in #531
  • Fix invalid RFC 3339 in TimeTagToAny (unreleased new feature) by @benluddy in #540
  • Refactor and improve code by @fxamacker in #542
  • Use "cbor:" prefixed error msg when decoding with non-default TimeTagToAnyMode setting by @fxamacker in #544
  • Disable conflicting encode options when marshaling cbor.Tag. by @benluddy in #546
  • Improve byte string format decoding options by @fxamacker in #550
  • Replace *errors.errorString with InadmissibleTagContentTypeError by @fxamacker in #552
  • Rename ByteSliceMode to ByteSliceLaterFormatMode, etc by @fxamacker in #554
  • Fix panic using SortFastShuffle (unreleased new feaure) to encode a struct with no fields. by @benluddy in #556

🐞 Bug Fixes

  • Check well-formedness of data from MarshalCBOR by @fxamacker in #485
  • Treat map keys matching the same struct field as duplicates. by @benluddy in #492
  • Decouple time.Time parsing from empty interface behavior. by @benluddy in #503

Other changes

Details

Full Changelog

Full Changelog: v2.6.0...v2.7.0

v2.7.0-rc.0

17 Jun 04:15
04bb1e7
Compare
Choose a tag to compare
v2.7.0-rc.0 Pre-release
Pre-release

What's Changed Since v2.6.0

v2.7.0 adds features and improvements that help large projects (e.g. Kubernetes) use CBOR as an alternative to JSON and Protocol Buffers. Other improvements include speedups, improved memory use, bug fixes, decoding/encoding options, etc.

Special thanks to @benluddy for contributing features, optimizations, improvements, bug fixes, and discussions! 👍

⭐ Features and Optimizations

  • Add decoding option TimeTagToAny to produce RFC3339 when decoding time into interface{} by @ssuriyan7 in #506
  • Bypass sorting overhead for single-entry maps. by @benluddy in #516
  • Add decode option to allow rejecting inputs that contain certain simple values. by @benluddy in #481
  • Allow rejection of NaN and Inf float values on encode and decode. by @benluddy in #513
  • Add SortMode to encode struct fields in a less predictable order. by @benluddy in #515
  • Add a decoding option to allow decoding byte string into time.Time. by @benluddy in #524
  • Encode structs directly to output buffer. by @benluddy in #519
  • Support automatic conversion between text and binary string representations by @benluddy in #476
  • Add a method for marshaling directly into a user-provided buffer. by @benluddy in #521
  • Add options to disable BinaryMarshaler/BinaryUnmarshaler support. by @benluddy in #526
  • Add option to reject decoding bignum tags and encoding big.Int. by @benluddy in #527
  • Improve speed & memory use for Diagnose() and DiagnoseFirst() by @benluddy in #533
  • Refactor sorted map encode to use fewer buffers for nested maps. by @benluddy in #537
  • Allow user to specify buffer by adding cbor.MarshalToBuffer(), UserBufferEncMode interface, etc. by @fxamacker in #553

🛠 Improvements

Improvements focused on fixing or updating new features. For bug fixes to features that were already released, see 🐞 Bug Fixes.

  • Refactor to reuse functions and improve code coverage by @fxamacker in #531
  • Fix invalid RFC 3339 in TimeTagToAny (unreleased new feature) by @benluddy in #540
  • Refactor and improve code by @fxamacker in #542
  • Use "cbor:" prefixed error msg when decoding with non-default TimeTagToAnyMode setting by @fxamacker in #544
  • Disable conflicting encode options when marshaling cbor.Tag. by @benluddy in #546
  • Improve byte string format decoding options by @fxamacker in #550
  • Replace *errors.errorString with InadmissibleTagContentTypeError by @fxamacker in #552
  • Rename ByteSliceMode to ByteSliceLaterFormatMode, etc by @fxamacker in #554
  • Fix panic using SortFastShuffle (unreleased new feaure) to encode a struct with no fields. by @benluddy in #556

🐞 Bug Fixes

  • Check well-formedness of data from MarshalCBOR by @fxamacker in #485
  • Treat map keys matching the same struct field as duplicates. by @benluddy in #492
  • Decouple time.Time parsing from empty interface behavior. by @benluddy in #503

Other changes

Details

Full Changelog

Full Changelog: v2.6.0...v2.7.0-rc.0

v2.7.0-beta

10 Jun 04:02
878cfef
Compare
Choose a tag to compare
v2.7.0-beta Pre-release
Pre-release

What's Changed

Latest changes (among many) include cbor.MarshalToBuffer() and UserBufferEncMode interface to support user specified buffer (alternative to built-in buffer pool). This was going to be a release candidate but recent improvements (June 9, 2024) needs fuzzing, etc.

Special thanks to @benluddy for contributing features, optimizations, improvements, bug fixes, and discussions! 👍

⭐ Features and Optimizations

  • Add decoding option TimeTagToAny to produce RFC3339 when decoding time into interface{} by @ssuriyan7 in #506
  • Bypass sorting overhead for single-entry maps. by @benluddy in #516
  • Add decode option to allow rejecting inputs that contain certain simple values. by @benluddy in #481
  • Allow rejection of NaN and Inf float values on encode and decode. by @benluddy in #513
  • Add SortMode to encode struct fields in a less predictable order. by @benluddy in #515
  • Add a decoding option to allow decoding byte string into time.Time. by @benluddy in #524
  • Encode structs directly to output buffer. by @benluddy in #519
  • Support automatic conversion between text and binary string representations by @benluddy in #476
  • Add a method for marshaling directly into a user-provided buffer. by @benluddy in #521
  • Add options to disable BinaryMarshaler/BinaryUnmarshaler support. by @benluddy in #526
  • Add option to reject decoding bignum tags and encoding big.Int. by @benluddy in #527
  • Improve speed & memory use for Diagnose() and DiagnoseFirst() by @benluddy in #533
  • Refactor sorted map encode to use fewer buffers for nested maps. by @benluddy in #537
  • Allow user to specify buffer by adding cbor.MarshalToBuffer(), UserBufferEncMode interface, etc. by @fxamacker in #553

🛠 Improvements

Improvements focused on fixing or updating new features. For bug fixes to features that were already released, see 🐞 Bug Fixes.

  • Refactor to reuse functions and improve code coverage by @fxamacker in #531
  • Fix invalid RFC 3339 in TimeTagToAny (unreleased new feature) by @benluddy in #540
  • Refactor and improve code by @fxamacker in #542
  • Use "cbor:" prefixed error msg when decoding with non-default TimeTagToAnyMode setting by @fxamacker in #544
  • Disable conflicting encode options when marshaling cbor.Tag. by @benluddy in #546
  • Improve byte string format decoding options by @fxamacker in #550
  • Replace *errors.errorString with InadmissibleTagContentTypeError by @fxamacker in #552
  • Rename ByteSliceMode to ByteSliceLaterFormatMode, etc by @fxamacker in #554

🐞 Bug Fixes

  • Check well-formedness of data from MarshalCBOR by @fxamacker in #485
  • Treat map keys matching the same struct field as duplicates. by @benluddy in #492
  • Decouple time.Time parsing from empty interface behavior. by @benluddy in #503

Other changes

Details

Full Changelog

Full Changelog: v2.6.0...v2.7.0-beta

v2.6.0

14 Feb 13:37
1f67e1e
Compare
Choose a tag to compare

This release adds important new features, optimizations, bug fixes, and 3 new contributors. Fuzz tests passed 5+ billion execs for CBOR encoding and decoding functions. Minimum officially supported Go is bumped to Go 1.17.

⭐ Features and Optimizations

  • Add FieldNameMatching decode option. by @benluddy in #433
  • Add option to decode CBOR bignum to interface{} as *big.Int by @fxamacker in #456
  • Add 2 more options for decoding CBOR integers to interface{} by @fxamacker in #457
  • Add encoding option to specify how omitempty fields are encoded by @dinhxuanvu in #453
  • New options for encoding Go strings to and from CBOR byte strings by @benluddy in #465
  • Add options to support byte string map keys as struct field names by @benluddy in #472
  • Add option to specify how to decode unrecognized CBOR tag to any by @ssuriyan7 in #475

🚀 Notable Optimizations

  • Improve memory allocs 🗜️ and speed 🚀 of encoding maps by using Go 1.18-1.20 features by @dinhxuanvu in #468

    📊 Benchmarks

    benchstat results provided by @dinhxuanvu

    Screenshot of CBOR benchmarks

    NOTE: Go 1.17 is the minimum version supported by v2.6.0. Go 1.20+ will use optimizations introduced by PR 468.

🛠 Improvements

🐞 Bug Fixes

  • Fix panic when decoding CBOR nil to *cbor.SimpleValue by @fxamacker in #461
  • Fix cbor.SimpleValue encoding and decoding by @fxamacker in #464
  • Fix EncOption/DecOption unset fields on mode regurgitation. by @benluddy in #480

📖 Docs and CI

New Contributors

Full Changelog: v2.5.0...v2.6.0

v2.6.0-rc.0

12 Feb 04:49
1f67e1e
Compare
Choose a tag to compare
v2.6.0-rc.0 Pre-release
Pre-release

⭐ Features and Optimizations

  • Add FieldNameMatching decode option. by @benluddy in #433
  • Add option to decode CBOR bignum to interface{} as *big.Int by @fxamacker in #456
  • Add 2 more options for decoding CBOR integers to interface{} by @fxamacker in #457
  • Add encoding option to specify how omitempty fields are encoded by @dinhxuanvu in #453
  • New options for encoding Go strings to and from CBOR byte strings by @benluddy in #465
  • Add options to support byte string map keys as struct field names by @benluddy in #472
  • Add option to specify how to decode unrecognized CBOR tag to any by @ssuriyan7 in #475

🚀 Notable Optimizations

  • Improve memory allocs 🗜️ and speed 🚀 of encoding maps by using Go 1.18-1.20 features by @dinhxuanvu in #468

    📊 Benchmarks

    benchstat results provided by @dinhxuanvu

    Screenshot of CBOR benchmarks

    NOTE: Go 1.17 is the minimum version supported by v2.6.0. Go 1.20+ will use optimizations introduced by PR 468.

🛠 Improvements

🐞 Bug Fixes

  • Fix panic when decoding CBOR nil to *cbor.SimpleValue by @fxamacker in #461
  • Fix cbor.SimpleValue encoding and decoding by @fxamacker in #464
  • Fix EncOption/DecOption unset fields on mode regurgitation. by @benluddy in #480

📖 Docs and CI

New Contributors

Full Changelog: v2.5.0...v2.6.0-rc.0

v2.5.0

14 Aug 03:42
3b32167
Compare
Choose a tag to compare

This release adds important new features, optimizations, bug fixes, and 8 new contributors. Fuzz tests passed 5+ billion execs for CBOR encoding and decoding functions.

Notable changes are split into 3 categories.

⭐ Notable Changes to Review Before Upgrading

These new features and bug fixes were cherry-picked to highlight for review (for projects using older version).

  • PR 370: Add SimpleValue type to more fully support CBOR Simple Values, including values not assigned by IANA and...
  • PR 376: Add ByteString type to support CBOR maps with byte string keys because Go doesn't allow []byte as map keys and...
  • PR 379: Make Decoder.Decode() return io.ErrUnexpectedEOF instead of io.EOF on EOF if current CBOR data item is incomplete.
  • PR 380: Make Unmarshal() and Valid() return cbor.ExtraneousDataError (instead of ignoring extraneous data if any remain).
  • PR 387: Retry in Decoder if io.Reader's Read() returns 0 bytes read with nil error. Add tests to get 100% code coverage for stream.go.

⭐ More Notable Changes

  • PR 342: Add DecOptions.UTF8 to decode invalid UTF-8. Default is unchanged (reject invalid UTF-8 and return error).
  • PR 355 Allow MaxNestedLevels setting to be configured up to 65535.
  • PR 352, 377: Add EncOptions.NilContainersMode to encode nil Go maps and slices as either CBOR nil (default) or empty container.
  • PR 381: Add Decoder.Skip() to skip CBOR data item in CBOR Sequences (RFC 8742).
  • PR 386 Add functions for Extended Diagnostic Notation (RFC 8610 Appendix G).
  • PR 398 Add UnmarshalFirst() to decode CBOR data item and also return remaining bytes.
  • PR 400 Deprecate Valid and add Wellformed to replace it.
  • PR 402 Add UnmarshalFirst() to DecMode interface.
  • PR 412 Add Decoder.Buffered to return remaining data in buffer.

🚀 Notable Changes Affecting Speed or Memory Use

  • PR 335: Reuse underlying array if RawMessage has sufficient capacity.
  • PR 382: Return buffer to pool in Encode(). It adds a bit of overhead to Encode() but NewEncoder().Encode() is faster and uses less memory.

Benchmark comparison for PR 382 shows significant improvement in speed and memory use.

🚀 (click to expand) 🚀

Benchmarks provided by @x448 for NewEncoder().Encode() optimized in PR 382.

$ benchstat bench-v2.4.0.log bench-f9e6291.log 
goos: linux
goarch: amd64
pkg: github.com/fxamacker/cbor/v2
cpu: 12th Gen Intel(R) Core(TM) i7-12700H
                                                     │ bench-v2.4.0.log │  bench-f9e6291.log                  │
                                                     │      sec/op      │   sec/op     vs base                │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                   236.70n ± 2%   58.04n ± 1%  -75.48% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20         238.00n ± 2%   63.93n ± 1%  -73.14% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20          238.65n ± 2%   64.88n ± 1%  -72.81% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20               242.00n ± 2%   63.00n ± 1%  -73.97% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20               245.60n ± 1%   68.55n ± 1%  -72.09% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                 243.20n ± 3%   68.39n ± 1%  -71.88% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                 563.0n ± 2%    378.3n ± 0%  -32.81% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20       2.043µ ± 2%    1.906µ ± 2%   -6.75% (p=0.000 n=10)
geomean                                                    349.7n         122.7n       -64.92%

                                                     │ bench-v2.4.0.log │    bench-f9e6291.log                │
                                                     │       B/op       │    B/op     vs base                 │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                     128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20           128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20            128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20                 128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20                 128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                   128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                   128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20         544.0 ± 0%   416.0 ± 0%   -23.53% (p=0.000 n=10)
geomean                                                      153.4                    ?                       ¹ ²
¹ summaries must be >0 to compute geomean
² ratios must be >0 to compute geomean

                                                     │ bench-v2.4.0.log │    bench-f9e6291.log                │
                                                     │    allocs/op     │ allocs/op   vs base                 │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                     2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20           2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20            2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20                 2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20                 2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                   2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                   2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20         28.00 ± 0%   26.00 ± 0%    -7.14% (p=0.000 n=10)
geomean                                                      2.782                    ?                       ¹ ²
¹ summaries must be >0 to compute geomean
² ratios must be >0 to compute geomean

What's Changed

All merged pull requests are listed here (including the already highlighted ones).

(click to expand)

Changes to Code

Most coding changes here were already mentioned.

  • Reuse underlying array if RawMessage has sufficient capacity by @zensh in #335
  • Add decoding option to allow invalid UTF-8 by @fxamacker in #342
  • Allow MaxNestedLevels to be up to 65535 by @immesys in #355
  • Add support for unassigned/reserved CBOR simple values by @fxamacker in #370
  • Add ByteString type to support any CBOR byte string by @fxamacker in #376
  • add option to enforce nil container marshaling as empty containers by @dedefer in #352
  • Refactor NilContainersMode option by @fxamacker in #377
  • Make Decoder.Decode() return io.UnexpectedEOF instead of io.EOF if CBOR data item is truncated by @fxamacker in #379
  • Fix handling of extra data in Unmarshal() & Valid() by @fxamacker in #380
  • Add Decoder.Skip() to skip CBOR data item in CBOR Sequences (RFC 8742) by @fxamacker in #381
  • Return buffer to pool when using Encoder by @fxamacker in #382
  • Retry in decoder if Read() returns 0 bytes read with nil error by @fxamacker in #387
  • Deprecate Valid() and add Wellformed() to replace it by @fxamacker in #400
  • Add UnmarshalFirst by @immesys in #398
  • Add UnmarshalFirst to DecMode interface by @fxamacker in #402
  • Add functions for Extended Diagnostic Notation (RFC 8610 Appendix G) by @zensh in #386
  • Add more tests for Diagnose and DiagnoseFirst by @fxamacker in #405
  • Fix Diagnose to return io.EOF error on empty data by @fxamacker in #410
  • Add Decoder.Buffered to return buffered data by @fxamacker in #412
  • Fix Diagnose() to return io.EOF error on empty data by @fxamacker in #410

Changes to CI, Comments, and Docs

Read more

v2.5.0-beta5

26 Jun 04:16
48c838c
Compare
Choose a tag to compare

v2.5.0-beta5 is fuzz tested and production quality, but docs need to be updated before releasing v2.5.0.
IMPORTANT: Please view these release notes (esp. v2.5.0-beta) before upgrading from v2.4 or earlier:

  • v2.5.0-beta4 - Bugfix for Diagnose to return io.EOF on empty data.
  • v2.5.0-beta3 - Add functions: Diagnose, DiagnoseFirst, UnmarshalFirst, Wellformed.
  • v2.5.0-beta2 - Bugfix to retry in Decoder if io.Reader's Read() returns 0 bytes read with nil error.
  • v2.5.0-beta - Notable improvements, optimizations, bugfixes, and 8 new contributors!

What's Changed since v2.5.0-beta4

Full Changelog: v2.5.0-beta4...v2.5.0-beta5

v2.5.0-beta4

18 Jun 05:40
340a5fb
Compare
Choose a tag to compare

Features already present in v2.4 are release quality. Newly added functions require more fuzzing and docs.
IMPORTANT: Please view these release notes (esp. v2.5.0-beta) before upgrading from v2.4:

  • v2.5.0-beta3 - Add Diagnose(), DiagnoseFirst(), UnmarshalFirst(), Wellformed().
  • v2.5.0-beta2 - Bugfix and release notes include benchmark comparison of v2.5.0-beta2 vs v2.4.0.
  • v2.5.0-beta - Notable improvements, optimizations, bugfixes, and 8 new contributors!

What's Changed Since v2.5.0-beta3

  • Bump safer-golangci-lint to 1.52.2 by @x448 in #408
  • Fix Diagnose() to return io.EOF error on empty data by @fxamacker in #410
  • Update README.md for cbor v2.5.0-beta4 by @fxamacker in #411

Full Changelog: v2.5.0-beta3...v2.5.0-beta4

v2.5.0-beta3

16 May 04:45
a52d9a5
Compare
Choose a tag to compare
v2.5.0-beta3 Pre-release
Pre-release

This release adds 4 new functions: UnmarshalFirst, Diagnose, DiagnoseFirst, and Wellformed. In addition Valid function is deprecated (use Wellformed instead).

  • UnmarshalFirst decodes first CBOR data item and returns remaining bytes. This is useful for CBOR Sequences (RFC 8742) and also because Unmarshal was fixed in v2.5.0-beta to return ExtraneousDataError if there are remaining bytes.
  • Diagnose and DiagnoseFirst returns human-readable Extended Diagnostic Notation (RFC 8610 Appendix G) for the given CBOR encoding. These are useful for debugging and allows more user-friendly error logging.

What's Changed (since v2.5.0-beta2)

  • Deprecate Valid() and add Wellformed() to replace it by @fxamacker in #400
  • Add UnmarshalFirst by @immesys in #398
  • Add UnmarshalFirst to DecMode interface by @fxamacker in #402
  • Add functions for Extended Diagnostic Notation (RFC 8610 Appendix G) by @zensh in #386
  • Add more tests for Diagnose and DiagnoseFirst by @fxamacker in #405

Changes to CI and Documentation

Special Thanks

Thank you @immesys, @zensh, and @x448 for your contributions!

Full Changelog: v2.5.0-beta2...v2.5.0-beta3

v2.5.0-beta2

21 Feb 00:02
f9e6291
Compare
Choose a tag to compare
v2.5.0-beta2 Pre-release
Pre-release

There are fewer changes in v2.5.0-beta2 compared to v2.5.0-beta. Speed and memory improvements since v2.4.0 are from v2.5.0-beta. Tests cover 98.5% and reached 100% coverage for stream.go (e.g. Decoder using the provided io.Reader).

What's Changed Since v2.5.0-beta

  • Retry in Decoder if io.Reader's Read() returns 0 bytes read with nil error by @fxamacker in #387
    • This PR also added more tests so we have 100% coverage for stream.go.
  • Update fxamacker_cbor_banner.png for v2.5 by @x448 in #383
  • bump safer-golangci-lint.yml to 1.51.1 by @x448 in #389

Go language recommends that Read implementations for io.Reader avoid returning 0 with nil error. It also recommends that callers receiving 0 and nil from Read should treat it as if nothing happened. This edge case is handled properly now.

@x448 shared benchmarks comparison of v2.4.0 vs v2.5.0-beta2. Performance improvements are from v2.5.0-beta.

Comparison of v2.5.0-beta2 vs v2.4.0

$ benchstat benchNewEncoderEncode-v2.4.0.log benchNewEncoderEncode-f9e6291.log 
goos: linux
goarch: amd64
pkg: github.com/fxamacker/cbor/v2
cpu: 12th Gen Intel(R) Core(TM) i7-12700H
                                                     │ benchNewEncoderEncode-v2.4.0.log │  benchNewEncoderEncode-f9e6291.log  │
                                                     │              sec/op              │   sec/op     vs base                │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                                   236.70n ± 2%   58.04n ± 1%  -75.48% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20                         238.00n ± 2%   63.93n ± 1%  -73.14% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20                          238.65n ± 2%   64.88n ± 1%  -72.81% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20                               242.00n ± 2%   63.00n ± 1%  -73.97% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20                               245.60n ± 1%   68.55n ± 1%  -72.09% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                                 243.20n ± 3%   68.39n ± 1%  -71.88% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                                  563.0n ± 2%   378.3n ± 0%  -32.81% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20                        2.043µ ± 2%   1.906µ ± 2%   -6.75% (p=0.000 n=10)
geomean                                                                     349.7n        122.7n       -64.92%

                                                     │ benchNewEncoderEncode-v2.4.0.log │    benchNewEncoderEncode-f9e6291.log    │
                                                     │               B/op               │    B/op     vs base                     │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                                     128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20                           128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20                            128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20                                 128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20                                 128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                                   128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                                   128.0 ± 0%     0.0 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20                         544.0 ± 0%   416.0 ± 0%   -23.53% (p=0.000 n=10)
geomean                                                                      153.4                    ?                       ¹ ²
¹ summaries must be >0 to compute geomean
² ratios must be >0 to compute geomean

                                                     │ benchNewEncoderEncode-v2.4.0.log │    benchNewEncoderEncode-f9e6291.log    │
                                                     │            allocs/op             │ allocs/op   vs base                     │
NewEncoderEncode/Go_bool_to_CBOR_bool-20                                     2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_uint64_to_CBOR_positive_int-20                           2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_int64_to_CBOR_negative_int-20                            2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_float64_to_CBOR_float-20                                 2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]uint8_to_CBOR_bytes-20                                 2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_string_to_CBOR_text-20                                   2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_[]int_to_CBOR_array-20                                   2.000 ± 0%   0.000 ± 0%  -100.00% (p=0.000 n=10)
NewEncoderEncode/Go_map[string]string_to_CBOR_map-20                         28.00 ± 0%   26.00 ± 0%    -7.14% (p=0.000 n=10)
geomean                                                                      2.782                    ?                       ¹ ²
¹ summaries must be >0 to compute geomean
² ratios must be >0 to compute geomean

Full Changelog: v2.5.0-beta...v2.5.0-beta2