klarna/avlizer

Avro Data Serializer for Erlang

avlizer - Avro Serializer for Erlang

avlizer is built on top of erlavro The design goal of this application is to provide APIs for users to serialize or deserialize data with minimal overhead of schema management.

Integrations

Confluent Schema Registry

Confluent schema registry provides REST APIs for schema registration and lookups.

Config

Make sure schema registry URL is present in sys.config as below

{avlizer, [{avlizer_confluent, #{schema_registry_url => URL}}]}

Or set os env variable:

AVLIZER_CONFLUENT_SCHEMAREGISTRY_URL

Authentication support

avlizer currently supports Basic authentication mechanism, to use it make sure schema_registry_auth tuple {Mechanism, File}, is present in sys.config, where File is the path to a text file which contains two lines, first line for username and second line for password

{avlizer, [{avlizer_confluent, #{
    schema_registry_url => URL, 
    schema_registry_auth => {basic, File}}
}]}

Or set authorization env variables:

AVLIZER_CONFLUENT_SCHEMAREGISTRY_AUTH_MECHANISM

AVLIZER_CONFLUENT_SCHEMAREGISTRY_AUTH_USERNAME

AVLIZER_CONFLUENT_SCHEMAREGISTRY_AUTH_PASSWORD

avlizer's Issues

Upgrade to erlavro 2.6.1

Hi, nice library! Could you upgrade to the latest version of erlavro?

FR: Allow passing decode options to avlizer_confluent:make_decoder & avlizer_confluent:decode

Hi!

The underlying erlavro library allows specifying codec_options for the decoder however avlizer has no way to pass those options to the decoder and defaults to [{encoding, avro_binary}]. I would like to propose extending avlizer_confluent:make_decoder/n and avlizer_confluent:decode/n API and allowing passing those options as that would allow for a more versatile usage of the library.

According to my initial observations that should not be a difficult thing to implement and given approval, I would volunteer for creating a PR

Please release 0.4.0 to hex.pm ❤️

Get encoder with schema id registry added in result

Currently, there is no possible way to get an encoder that can encode the pass object and add schema id together. If we had to do it currently it's 2 steps.

1. encode function to encode data
2. tag_data function to append the schema id.

Purposed approach
Merge these 2 steps into 1 and return only 1 encoder. Which client can use directly.

PS: I would like to raise PR with this feature.