• Fix outstanding rchk errors
  • Prepare for CRAN
  • Fix some compilation warnings.
  • Update documentation.
  • Update to yyjson v0.9.0
  • Bug fix for serializing NA in factors
  • Re-introduce GeoJSON support
  • Increase maximum number of allowed data.frame columns (during parsing) to 2048
  • New fast_numerics flag when writing.
    • Default FALSE
    • If TRUE the user is guaranteeing that there are no NA, NaN or Inf values in the numeric and integer vectors, and thus a faster method for writing these vectors to JSON can be used.
  • Changed writing of raw vectors to always use the fast_numerics method, as raw R vectors (by definition) will not have NA, NaN or Inf values.
  • Fixes for CRAN
    • Adjust pointer arithmetic when calling output_verbose_error() to avoid overflow of size_t
  • Fixes for CRAN
    • Fixed warnings when building with MAKEVARS PKG_CFLAGS = -Wconversion
    • Fixed C error where raw buffer was passed to strlen() leading to a check failure on cran with clang-ASAN
  • Fixes for CRAN
    • DESCRIPTION fix: Write C as 'C'
    • Add link to original yyjson library
    • Fix name in LICENSE
  • Fixes for CRAN
    • Simplify example to remove checkRd NOTE
    • Platform specific handling of error location format string to fix WARNING
  • Remove NDJSON and GeoJSON code to simplify preparation for CRAN. Will re-introduce this code in future releases.
  • Updated to YYJSON v0.8.0
  • Fix an off-by-one error when reporting line numbers in NDJSON handling.
  • Increase buffer size when reading lines from NDJSON files.
    • MAX_LINE_LENGTH now 131072 (was 10000)
  • Writing to JSON objects now supports a digits argument for rounding floating point values to the specified number of significant digits
    • digits = -1 means don’t do any rounding
    • digits = 0 rounds floating point values to integers (and writes the values as JSON integers)
  • Refactored options for simplification to data.frame
    • removed vectors_to_df
    • replaced with obj_of_arrs_to_df
    • added arr_of_obs_to_df
  • length-1 vectors marked with class AsIs (using a call to I()) will never be unboxed i.e. will always be serialized as a JSON []-array with one element.
  • Added parse option length1_array_asis. If TRUE then automatically add the class AsIs to the object object.
  • Added pre-calculation and caching of data.frame column types for faster serialization when outputting row-by-row.
  • Unifed naming scheme
  • FEATURE: Added promote_num_to_string in opts_read_json() to enable forced promotion of numerics to string
  • BUGFIX: fixes for handling of geometry collection when reading and writing.
  • TESTING: More tests included for output to geojson
  • TESTING: Refactored testing of ‘sf’ objects
  • Bug fix for checking attributes on list columns. This surfaced more errors to do with setting XYZ geometry type.
  • More extensive testing of write_geojson_str()
  • Update to option setting to remove overhead
  • Added read_json_raw() to parse JSON directly from a raw vector (which contains UTF-8 string data)
  • Initial release