Further performance optimisations. This yields an approx 10% improvement in decoding performance where the schema contains arrays and/or maps.

Performance improvements

Reuse of Avro encoder/decoder
The various encoder/decoder types (binary, json, pretty json) are now created when you compile the schema (and referenced by the same foreign) and these are then reused for each subsequent encode or decode. This yields an approx 50% improvement in encoding and decoding performance.
One limitation of this is that Avro encoders/decoders do not support concurrent access. Therefore, the approach of reusing the encoder/decoder won’t work with peach. In order to support such usage, a new option MULTITHREADED has been added which reverts back to the original behaviour of creating a new encoder/decoder on the fly. If you want to use encode or decode with peach this option must be set to non-zero.

Elimination of unnessary copies
Handling of the serialised buffers has been optimised to eliminate unnessary copies. This yields an approx 15% improvement in encoding and decoding performance.

Build linux release package on Ubuntu 14 (Trusty) rather than Ubuntu 16 (Xenial) for better compatibility with older distros such as RHEL/Centos 7.

Build linux release package on Ubuntu 16 (Xenial) rather than Ubuntu 18 (Bionic) for better compatibility with older distros such as RHEL/Centos 7.

Apache Avro is a data serialization system which provides:

• Rich data structures.
• A compact, fast, binary data format.

Avrokdb supports the use of Avro in kdb+ to:

• Compile an Avro JSON schema.
• Encode a kdb+ object to Avro using a schema.
• Decode from Avro to a kdb+ object using a schema.

It support all the Avro datatypes and logical types.