AvroCodec

Binary-decode bytes under schema to a generic record — no typed codec involved. Catches apache-avro's binary-decoder throws (IOException / EOFException / AvroRuntimeException) into AvroFailure.BinaryParseFailed.

Binary-decode bytes (encoded under readSchema) and RESOLVE them into writeSchema via apache-avro's schema resolution (reorder / default / promotion / aliases) — the drift counterpart of decodeRecord (which assumes one schema). Named from the reader's point of view: readSchema is the schema the bytes are read in, writeSchema the shape they resolve into (what you'd write). In apache-avro's own vocabulary these are the writer schema and the reader schema respectively — hence GenericDatumReader(readSchema, writeSchema) below. Incompatible schemas → AvroFailure.ResolveFailed.

decodeResolvedRecord resolving readSchema → codec's schema (the write schema), then the codec's Any ⇒ A side — the drift counterpart of decodeValue.

decodeRecord under codec's schema, then the codec's Any ⇒ A side — AvroFailure.DecodeFailed when the parsed record doesn't line up with A.

Binary-encode an already-Any-shaped datum under schema — the write-side counterpart of decodeRecord. GenericDatumWriter throws AvroTypeException / NullPointerException / ClassCastException when the datum doesn't line up with the schema; caught into AvroFailure.EncodeFailed.

The codec's A ⇒ Any side, then encodeRecord under codec's schema.

Default project-wide kindlings config: identity field-name transform, no namespace override, no decimal config. Users can shadow this with their own given AvroConfig if they need a snake-cased schema or a forced namespace.

Combine kindlings' derived AvroEncoder[A] / AvroDecoder[A] / AvroSchemaFor[A] into a single AvroCodec. Each kindlings typeclass is summoned independently, so this works whether the user has hand-written one of the sides or used kindlings' macro derivation for both.