git.iko.soy/typeable/aeson
on 0.8.0.1 help sign in
typeable/aeson/changelog.md
129 lines 4.3 kB history raw
10.8.0.1
3* Support time-1.5.
50.8.0.0
7* Add ToJSON and FromJSON instances for tuples of up to 15 elements.
90.7.1.0
11* Major compiler and library compatibility changes: we have dropped
12 support for GHC older than 7.4, text older than 1.1, and bytestring
13 older than 0.10.4.0. Supporting the older versions had become
14 increasingly difficult, to the point where it was no longer worth
15 it.
170.7.0.0
19* The performance of encoding to and decoding of bytestrings have both
20 improved by up to 2x, while also using less memory.
22* New dependency: the scientific package lets us parse floating point
23 numbers more quickly and accurately.
25* eitherDecode, decodeStrictWith: fixed bugs.
27* Added FromJSON and ToJSON instances for Tree and Scientific.
29* Fixed the ToJSON instances for UTCTime and ZonedTime.
310.6 series
33* Much improved documentation.
35* Angle brackets are now escaped in JSON strings, to help avoid XSS
36 attacks.
38* Fixed up handling of nullary constructors when using generic
39 encoding.
41* Added ToJSON/FromJSON instances for:
43 * The Fixed class
44 * ISO-8601 dates: UTCTime, ZonedTime, and TimeZone
46* Added accessor functions for inspecting Values.
48* Added eitherDecode function that returns an error message if
49 decoding fails.
510.5 to 0.6
53* This release introduces a slightly obscure, but
54 backwards-incompatible, change.
56 In the generic APIs of versions 0.4 and 0.5, fields whose names
57 began with a "_" character would have this character removed. This
58 no longer occurs, as it was both buggy and surprising
59 (https://github.com/bos/aeson/issues/53).
61* Fixed a bug in generic decoding of nullary constructors
62 (https://github.com/bos/aeson/issues/62).
640.4 to 0.5
66* When used with the UTF-8 encoding performance improvements
67 introduced in version 0.11.1.12 of the text package, this release
68 improves aeson's JSON encoding performance by 33% relative to aeson
69 0.4.
71 As part of achieving this improvement, an API change was necessary.
72 The fromValue function in the Data.Aeson.Encode module now uses the
73 text package's Builder type instead of the blaze-builder package's
74 Builder type.
760.3 to 0.4
78* The new decode function complements the longstanding encode
79 function, and makes the API simpler.
81* New examples make it easier to learn to use the package
82 (https://github.com/bos/aeson/tree/master/examples).
84* Generics support
86 aeson's support for data-type generic programming makes it possible
87 to use JSON encodings of most data types without writing any
88 boilerplate instances.
90 Thanks to Bas Van Dijk, aeson now supports the two major schemes for
91 doing datatype-generic programming:
93 * the modern mechanism, built into GHC itself
94 (http://www.haskell.org/ghc/docs/latest/html/users_guide/generic-programming.html)
96 * the older mechanism, based on SYB (aka "scrap your
97 boilerplate")
99 The modern GHC-based generic mechanism is fast and terse: in fact,
100 its performance is generally comparable in performance to
101 hand-written and TH-derived ToJSON and FromJSON instances. To see
102 how to use GHC generics, refer to examples/Generic.hs.
104 The SYB-based generics support lives in Data.Aeson.Generic and is
105 provided mainly for users of GHC older than 7.2. SYB is far slower
106 (by about 10x) than the more modern generic mechanism. To see how
107 to use SYB generics, refer to examples/GenericSYB.hs.
109* We switched the intermediate representation of JSON objects from
110 Data.Map to Data.HashMap which has improved type conversion
111 performance.
113* Instances of ToJSON and FromJSON for tuples are between 45% and 70%
114 faster than in 0.3.
116* Evaluation control
118 This version of aeson makes explicit the decoupling between
119 *identifying* an element of a JSON document and *converting* it to
120 Haskell. See the Data.Aeson.Parser documentation for details.
122 The normal aeson decode function performs identification strictly,
123 but defers conversion until needed. This can result in improved
124 performance (e.g. if the results of some conversions are never
125 needed), but at a cost in increased memory consumption.
127 The new decode' function performs identification and conversion
128 immediately. This incurs an up-front cost in CPU cycles, but
129 reduces reduce memory consumption.