changelog.md

• 2.11.2

  • Adjust licenses tag in .app.src due to hex changes.

• 2.11.1

  • avro_idl: the default IDL read_fun now confines imports to a root directory. Absolute paths and relative paths that escape the root through .. are refused with {error, {import_outside_root, Path}}, but .. is allowed as long as the resolved path stays under the root. The root defaults to the top-level caller's Cwd and can be overridden via the new {rootdir, Dir} option, supported by avro_idl:decode_schema/3, avro_idl:new_context/2, avro_idl:str_to_avpr/3, and the new avro_schema_store:new/3 / import_file/3 / import_files/3 entry points. Callers that need different resolution semantics (loading from arbitrary directories or from an in-memory store) can still supply their own function via the existing {read_fun, Fun} option, which bypasses path checks.

• 2.11.0

  • Deleted jsone as dependency. When OTP release is 27 or later, the default JSON provider module is json, otherwise jsone. For OTP release 26 or earlier version, you must add jsone-1.8.1 or newer in your project dependency. For OTP release 27 or later, you can choose to continue using jsone by calling avro:set_json_provider(jsone).

• 2.10.3

  • Allow union type to have zero member types.

• 2.10.2

  • Fix bytes and fixed JSON value decode

• 2.10.1

  • Fix dialyzer error.

• 2.10.0

  • Add map as avro store, and use it as default.
  • Changed to store type aliases as type's full name index, so the type store map (or dict) is less bloated.

• 2.9.10

  • Optimize avro:is_same_type/2
  • Upgrade jsone to 1.8.1

• 2.9.9

  • Enable builds with gcc-13
  • Improve @aliases support

• 2.9.8

  • Removed support for Rebar 2

• 2.9.7

  • Export OCF make_header and make_block functions

• 2.9.6

  • Add snappy compression for OCF

• 2.9.5

  • Removed workarounds for OTP < 21
  • Add map() to the avro:out() union type

• 2.9.4

  • Dropped ?MODULE from required error in erlavro.hrl to allow elixir to extract record details

• 2.9.3

  • Add OCF decoder hook
  • Dialyzer fixes

• 2.9.2

  • Move test files from priv to test/data

• 2.9.1

  • Add true and false to the atom exceptions when converting to strings

• 2.9.0

  • Encode atom values for string types when not nil or null

• 2.8.3

  • Allow arbitrary atoms in record_opt_name()

• 2.8.2

  • Encode atom nil as "null" for Elixir consumers

• 2.8.1

  • Support 'object' as custom type properties

• 2.8.0

  • Improve varint encoding performance
  • Support erlang map for avro record and map type in encoder
  • Allow decoder to decode avro record and map type as erlang map based on record_type and map_type options.

• 2.7.1

  • Issue #81 allow missing codec in ocf

• 2.7.0

  • Add schema compatibility check avro:is_compatible/2

• 2.6.5

  • Refine macros for get_stacktrace deprecation

• 2.6.4

  • Do not conditionally export functions

• 2.6.3

  • Upgrade jsone from 1.4.3 to 1.4.6 (for OTP-21)

• 2.6.2

  • Support Erlang 21.0 stacktraces

• 2.6.1

  • Allow "null" string as default value for 'null' type in union type record fields (contributer bka9)

• 2.6.0

  • Support Parsing Canonical Form for Schemas (contributer congini/@reachfh)

• 2.5.0

  • Add an Erlang implementation of the Avro CRC 64 fingerprint algorithm, avro:crc64_fingerprint/1.
  • Add avro:make_simple_decoder/2 and avro:make_simple_encoder/2 to simplify most common use cases. avro:make_decoder/2 and avro:make_encoder/2 are not deprecated and can still be used when there is a need to use the same decoer/encoder for multiple schema.
  • Remove fullname-derived-namespace = namespace-in-type-attributes assertition (#60)
  • Add avro_ocf:make_ocf/2 to allow building ocf content in memory

• 2.4.0

  • Support dict schema store, avro:make_decoder and avro:make_encoder accepts JSON directly
  • Fix default value encode/decode and validation Encoder now makes use of record field default value in case the field is mssing in input
  • Type reference validation options for avro:decode_schema/2
  • Changed default build from rebar to rebar3