Change Log

All notable changes to this project will be documented in this file, which follows the conventions of keepachangelog.com. This project adheres to Semantic Versioning.

Unreleased

...

1.1.0 - 2020-06-14

Added

• The base logic in the codec will write any collection supporting the standard Java List, Map, and Set interfaces as the corresponding CBOR types. Previously, this only worked directly on Clojure's collection types.
• Local calendar dates are supported using the proposed tags 100 (epoch) and 1004 (string), mapped to the java.time.LocalDate type.

1.0.0 - 2020-04-20

Added

• Errors about unhandled types now contain more type information about the value that could not be encoded. #15
• The clj-cbor.core/dispatch-superclasses function can be used to support inheritance-based write handler selection. #12

0.7.2 - 2019-06-28

Added

• Add clj-async-profiler support in the benchmarking code to produce flame graphs for finding hotspots.

Removed

• Codecs no longer support a configurable set tag; it is now fixed at 258 matching the IANA registry.

Fixed

• decode will not try to coerce inputs which are already input streams anymore. This fixes incremental/lazy decoding behavior. #13
• encode will not try to coerce outputs to an output stream anymore. Now an exception is thrown when a bad argument is given. This prevents lost writes when the final buffer is not flushed correctly. #14

0.7.1 - 2019-01-10

This is a performance-focused release that introduced a number of changes to speed up decoding performance. The main change is the switch to a jump table for the initial header byte. #9 #11

For a representative data sample of about ~97 KB, this brings the benchmarked decoding time from 8.454 ms to 4.089 ms, or about twice as fast!

Changed

• Upgrade to Clojure 1.10.0.
• Many operations in the codec are type-hinted to use primitive operations where possible.
• CBOR arrays and maps are built up using transients for performance.
• Decoding logic now uses a jump table.

Fixed

• A tagged-value major type with a streaming info code now results in a ::codec/illegal-stream error.

0.6.0 - 2017-12-23

Changed

• Upgrade to Clojure 1.9.0.

0.5.0 - 2017-11-08

This release fixes two of the longer-standing quirks with the library, which were unfortunately breaking changes. The fixes should be straightforward:

• Replace any (cbor/decode ...) with (cbor/decode-seq ...).
• Replace any (first (cbor/decode-seq ...)) with (cbor/decode ...).

If you have existing encoded data containing sets, you can use the following function to rewrite it:

(defn rewrite-cbor-sets
  [codec source dest]
  (with-open [input (io/input-stream source)
              output (io/output-stream dest)]
    (->>
      input
      (cbor/decode-seq (assoc codec :set-tag 13))
      (cbor/encode-seq codec output))))

If rewriting isn't an option, you can support reading sets via tag 13 by using a custom read handler:

(def compatible-codec
  (assoc-in cbor/default-codec [:read-handlers 13] set))

Changed

• Breaking: the default set tag is now 258, matching the IANA registry. #6
• Breaking: clj-cbor.core/decode now only decodes a single value; previous behavior moved to decode-seq. #7

Added

• clj-cbor.core/encode-seq writes a sequence of values to a byte array or output stream.
• clj-cbor.core/spit-all similarly writes a sequence of values to an output file like repeated calls to spit with :append true.

0.4.1 - 2017-05-17

Fixed

• Resolved an overflow issue when encoding Long/MIN_VALUE.
• Integer values are always decoded as Long values, no matter their encoded size. Previously, numbers between Short/MAX_VALUE and Integer/MAX_VALUE would return Integer values. #4 #5

0.4.0 - 2017-03-14

Added

• Implemented canonical mode sorting of map keys and set entries.

0.3.0 - 2016-01-05

Added

• Support self-describe CBOR tag 55799. This provides a 'magic' three-byte sequence to simplify detection of the CBOR format. The clj-cbor.core/self-describe function will wrap the given value with this tag.
• Tag codes have been factored out into integer constants where appropriate to improve consistency.
• Add spit, slurp, and slurp-all utility functions to the core namespace.

Changed

• Read handler functions are no longer called with the tag. This greatly simplifies their implementation and allows for reuse of existing transformation functions as-is.
• Error type :clj-cbor.codec/illegal-chunk renamed to illegal-chunk-type.

Fixed

• The decode function now properly raises an error when the input ends mid-value rather than at a top-level value boundary.

0.2.0 - 2016-12-28

This release includes 100% test coverage!

Added

• UUIDs are supported in binary form using tag 37.
• CBOR error keywords are organized into a hierarchy underneath :clj-cbor.error/encoding-error and :clj-cbor.error/decoding-error.

Changed

• clj-cbor.data.model renamed to clj-cbor.data.core.
• The clj-cbor.float16 functions from-bits and to-bits renamed to decode and encode, respectively.
• Undefined and SimpleValue records moved to new clj-cbor.data.simple namespace.
• TaggedValue record moved to new clj-cbor.data.tagged namespace.
• clj-cbor.header/write-major-int renamed to write.
• clj-cbor.header/read-size renamed to read-code.

Fixed

• Generally tighted up namespaces and reduced linkage where possible.

0.1.0 - 2016-12-23

Initial project release.