blob: ca678c970eb1675acad7dfa2ad4e850016821203 [file] [log] [blame] [edit]
luajson v0.10 Release Notes
===========================
User Visible Changes
--------------------
This release changes the system quite a bit from previous versions in
that there is more configurability. There still is quite a bit of strong
defaulting to make it accept more data formats and encode such that the
most strict decoder *should* decode it.
It is now possible to obtain wrapped-up encoders and decoders with options
"compiled" in. With decoders this is particularly bound-up (LPEG patterns),
encoders take advantage of this with less closure-count.
Extended features of newer LPEG versions can also be used as they are
available, either via function detection or version # parsing.
New decoder features:
* Single-quoted strings
* Function-call decoding
* Generic string management
* String post-processing option
* UTF-16 -> UTF-8 code decoding for "\uXXXX" from Emil Renner Berthing by default (not req)
* Handle extended UTF-8 spacing characters and BOF for inter-element spacing
New encoder features:
* Function-call encoding
* String pre-processing option
* Fully modular encoding system
* Encoding with similar option system as decoding
Plans for next release
----------------------
0.11
* UTF-8 encoding/decoding validation (currently decoding only supported)
* Optional \xXX encoding
* JSON encoding with fully 'strict' options
* Generic output mechanism (ropes, IO, etc)
1.0 beta
* Full API documentation
* Stable encoder/decoder option-apis. Final 1.0 release will maintain
that all options used in code will be future-proof, at least until 2.0
Updates since 0.9.1
===================
Emil Renner Berthing (1):
decoder:
Added proper utf8 decoding of strings
Thomas Harning Jr (38):
-ungrouped-
Merge branch '0.9-rockspec'
base:
Setup 0.9.1 rockspec with md5 of luaforge release
added luarocks module to check out luajson from git (requires git patch)
Preparing for next release pre-emptively
decode-util:
during a table merge, if a nil is located, skip it
decoder/test:
Added preliminary function-call handling
Require name is string and func is function
Support both patterns and strings as function name specifiers
function calls receive as first argument the name used
added support for multiple arguments to functions
Moved function-call decoder to separate module
added string post-processing utility
decoder/tests:
adds support for single-quoted strings w/ expected escape-handling
support unicode whitespaces as whitespace around values
decoder:
Fully modularized strings/number/calls data-types
Refactored boolean/null/undefined out into the 'others' module
Provide workaround for missing lpeg.type in lpeg 0.6, 0.7, ...
More stabilization of string decoder configuration as well as hooking unicode-handling in
optimized 'number' option-defaulting
unified configuration system into a fast-cached Tokend which only permits card-readng ops (writing in seprate referece...)
Fixed return assertion to prevent 'Invalid JSON data' from being returned as second result
Added simple LPEG version parser "x.y.z.." => x.y
Added support for LPEG 0.9 accumulator replacement (fold) which may be faster
string.char => string_char for optimization in utf8 decoder
encoder/decoder:
added support for 'undefined' value
encoder/test:
exposed and test string preprocessing option
encoder:
add string preprocessor handler (for encoding->utf8 mgmt)
Break apart the encoder into distinct modules + add call encoding
split out number encoder into own file to mirror decoder
added applicable configuration options from the decoder to the number encoder
enhanced configuration support to match the decoder's optionset where appropriate
added 'initialObject' check to assert that the root object is in fact an object/array
mirrored decoder's getDecoder => getEncoder setup
performed full modularization of the encoder + added function-encode tests
tests:
Removed package.path alteration since 'make check' handles that as well as luarocks
applied configuration optimizations take advantage of the encoder optimizations and simplifications from the decoders
added UTF-16 => UTF-8 encoder test
extends UTF tests with boundary tests and test # for break detect