Some datasets have nullable types in the schema, but as a consumer I don't care to distinguish between a null and the zero value. hamba/avro won't decode nullable types into a struct which does not contain pointer types, but it would be nice to allow this either on a per-field basis or else for the entire type.
I implemented a decode flag in a branch to allow decoding to non-pointer fields. Is there interest in adding something like this to hamba/avro? arg0net@4e73b21