Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | change :validate_utf8=false to still emit utf8lcg/utf8-string-replacement | Lamont Granquist | 2015-04-23 | 1 | -2/+3 |
| | | | | | | | | | | | validate_utf8=true raises on bad input validate_utf8=false now does not raise, but still produces valid UTF-8 clean JSON by replacing characters. the previous behavior of emitting bad JSON that was not UTF-8 clean when validate_utf8 was false is dropped (and this was behavior only of the C extension and not the ffi extension since the ffi layer must do some scrubbing of its own). | ||||
* | emit token that failed utf-8 validation | Lamont Granquist | 2015-04-21 | 2 | -33/+35 |
| | | | | plus code cleanup of the c-extension | ||||
* | bumping version to 2.1.02.1.0 | Lamont Granquist | 2015-04-20 | 1 | -1/+1 |
| | |||||
* | Merge pull request #19 from chef/lcg/stringio-encoding | Lamont Granquist | 2015-04-20 | 1 | -0/+9 |
|\ | | | | | support encoding StringIOs | ||||
| * | fix ffi side | Lamont Granquist | 2015-04-17 | 1 | -0/+9 |
| | | |||||
* | | add :unique_key_checking flag to parserlcg/unique_key_checking | Lamont Granquist | 2015-04-17 | 2 | -0/+5 |
|/ | | | | | can be used to error out if keys are duplicated in input rather than silently replacing. | ||||
* | add copyright notices | Lamont Granquist | 2015-04-14 | 10 | -0/+214 |
| | |||||
* | limit rescue to just LoadError | Lamont Granquist | 2015-04-13 | 1 | -2/+12 |
| | |||||
* | remove commented code | Lamont Granquist | 2015-04-06 | 1 | -4/+0 |
| | |||||
* | add comments | Lamont Granquist | 2015-04-06 | 1 | -0/+50 |
| | |||||
* | move dlopen to c extension | Lamont Granquist | 2015-04-06 | 2 | -5/+2 |
| | | | | jruby can't load this | ||||
* | fixing libname guessing logic | Lamont Granquist | 2015-04-06 | 3 | -43/+46 |
| | | | | | run through multiple different possible library names and extensions, favoring the libyajl2 gem versions first. | ||||
* | add native dlopen extension code | Lamont Granquist | 2015-04-06 | 2 | -51/+8 |
| | | | | | This will get dlopen in a consistent place on all the platforms which need it. | ||||
* | warn on stderrlcg/ffi-less-ffi-yajl | Lamont Granquist | 2015-03-24 | 1 | -2/+2 |
| | |||||
* | Remove ffi gem as a hard dependency | Lamont Granquist | 2015-03-24 | 3 | -16/+52 |
| | | | | | | | | | | | Copies the algorithm out of FFI.map_library_name into the FFI_Yajl#map_library_name function so that when we're loading up the C extension we don't need to also slurp in the ffi gem just to run that. The ffi gem is moved to a hard dependency of only the jruby version of the gem, and its a development dependency for the other ruby VMs (so rake spec will work against the ffi library), but its an optional library to install to make the ffi library work on VMs that support C extensions (MRI, RBX). | ||||
* | bumping version to 2.0.0 for release2.0.0 | Lamont Granquist | 2015-03-17 | 1 | -1/+1 |
| | |||||
* | Removing JSON gem compatibility layer | Lamont Granquist | 2015-03-17 | 1 | -140/+0 |
| | | | | | | | | | | | | | This causes issues because JSON and ffi-yajl get into fights over who monkeypatched the classes last and this causes random issues with whitespace disagreement between the JSON gem and the underlying libyajl C library. For deterministic output its better not to have fighting monkeypatches, it generates considerable chaos. If anyone wants to extract the json_gem.rb file into a gem of its own and support these monkeypatches going forwards that would be fine (there is, in fact, literally nothing that could be done to stop you), but we do not want to support these. | ||||
* | bump version to 1.4.0 | Lamont Granquist | 2015-02-17 | 1 | -1/+1 |
| | |||||
* | mixin windows? helper directly into the object | Lamont Granquist | 2015-02-17 | 2 | -5/+9 |
| | | | | | its useful to be able to include/extend the platform class and then use this sugar directly | ||||
* | Move #windows? to lib to allow not checking libname to determine platform | Tyler Vann-Campbell | 2015-02-17 | 2 | -1/+7 |
| | |||||
* | remove unnecessary linelcg/invalid-utf8 | Lamont Granquist | 2015-01-22 | 1 | -1/+0 |
| | |||||
* | support turning off validate_utf8 | Lamont Granquist | 2015-01-22 | 1 | -0/+1 |
| | |||||
* | catch and raise invalid string error better | Lamont Granquist | 2015-01-22 | 1 | -0/+2 |
| | |||||
* | include status code for Unknown Error | Elan Ruusamäe | 2015-01-22 | 1 | -1/+1 |
| | |||||
* | bump version to 1.3.1 | Lamont Granquist | 2014-11-24 | 1 | -1/+1 |
| | |||||
* | allow arbitrary Objects as keys | Lamont Granquist | 2014-11-24 | 1 | -1/+1 |
| | | | | i fixed Hashes/Arrays/true/false/nil, but forgot Object. | ||||
* | version bump to 1.3.01.3.0 | Lamont Granquist | 2014-11-22 | 1 | -1/+1 |
| | |||||
* | support calling #to_s on all types for keys | Lamont Granquist | 2014-11-22 | 1 | -26/+61 |
| | | | | | | compatibility with JSON gem and yajl-ruby closes #14 | ||||
* | fix bare object parsing | Lamont Granquist | 2014-11-22 | 1 | -5/+2 |
| | | | | closes #2 and #16 | ||||
* | fix #15 by requiring string or IO | Lamont Granquist | 2014-11-22 | 1 | -0/+2 |
| | |||||
* | Merge pull request #29 from opscode/lcg/better-ext-vs-ffi-logic | Lamont Granquist | 2014-11-22 | 1 | -2/+22 |
|\ | | | | | add better ext-vs-ffi logic | ||||
| * | line wrap on commentlcg/better-ext-vs-ffi-logic | Lamont Granquist | 2014-10-13 | 1 | -5/+8 |
| | | |||||
| * | add better ext-vs-ffi logic | Lamont Granquist | 2014-10-13 | 1 | -2/+19 |
| | | | | | | | | | | plus add warnings so its obvious what is happening and spammy warns will be a good indication of bugs. | ||||
* | | fix minor typo | Elan Ruusamäe | 2014-11-12 | 1 | -1/+1 |
|/ | |||||
* | add deprecation warning | Lamont Granquist | 2014-10-09 | 1 | -0/+2 |
| | |||||
* | the c ext loads and still tests fine on 1.8.7 | Lamont Granquist | 2014-10-09 | 1 | -2/+1 |
| | |||||
* | version bump to 1.2.01.2.0 | Lamont Granquist | 2014-10-09 | 1 | -1/+1 |
| | |||||
* | Merge pull request #27 from opscode/lcg/windows-libname | Lamont Granquist | 2014-10-09 | 2 | -0/+4 |
|\ | | | | | fix libnames for windows | ||||
| * | fix libnames for windowslcg/windows-libname | Lamont Granquist | 2014-10-08 | 2 | -0/+4 |
| | | | | | | | | this fixes the perf issues on windows where we were falling back to ffi | ||||
* | | Cleaning up code - letting yajl do more heavy lifting for us. Also removing ↵ | tyler-ball | 2014-10-07 | 2 | -3/+6 |
| | | | | | | | | version change since that will occur during release. | ||||
* | | Making necessary C changes to only call to_json if it is present | tyler-ball | 2014-10-07 | 1 | -1/+1 |
| | | |||||
* | | Upping version | tyler-ball | 2014-10-06 | 1 | -1/+1 |
| | | |||||
* | | If an object does not have .to_json, we no longer try to call it | tyler-ball | 2014-10-06 | 1 | -1/+5 |
|/ | |||||
* | use RbConfig instead of Config | Lamont Granquist | 2014-08-29 | 1 | -1/+1 |
| | |||||
* | bumping version to 1.1.01.1.0 | Lamont Granquist | 2014-08-26 | 1 | -1/+1 |
| | |||||
* | fix rbx using dlopen via FFI | Lamont Granquist | 2014-08-23 | 1 | -11/+44 |
| | | | | this means rbx uses FFI in ways that I cannot work around. | ||||
* | fix ruby 1.8.7 tests | Lamont Granquist | 2014-08-23 | 1 | -1/+1 |
| | |||||
* | add datetime encoding | Lamont Granquist | 2014-08-23 | 2 | -2/+45 |
| | | | | | | | | necessary for ohai json_gem spec complies with JSON gem format. this also arguably fixes issues with yajl-ruby where it does not encode date objects correctly. | ||||
* | bumping version + changelog for releas1.0.2 | Lamont Granquist | 2014-08-09 | 1 | -1/+1 |
| | |||||
* | ensure opts hash is never nil | Lamont Granquist | 2014-08-08 | 1 | -0/+1 |
| |