I'm working on a little library to wrap domain objects into Rows (using RowWithGetters) and am trying to implement a new LogicalType. I'm struggling to understand how it's supposed to interact with RowWithGetters however.
Let's say that my LogicalType has a java type of MyLogicalType.Value, and base type of byte[]. This means my I have something like: class MyLogicalType extends Schema.LogicalType<MyLogicalType.Value, byte[]> Importantly here, I need to implement a toInputType function that takes a byte[] and returns MyLogicalType.Value. What I'm trying to figure out is, if I construct a RowWithGetters, it seems like my getter for the logical type field needs to first convert from my domain value to a byte[], which will next be converted from a byte[] -> MyLogicalType.Value (in toInputType), and then (in the context of the SchemaCoder), be one more time converted from MyLogicalType.Value -> byte[] (in toBaseType). This seems like there's an extra round trip in here that there shouldn't be. For reference, the first part takes place in RowWithGetters.getValue [1] More confusingly, OneOfType seems to be special cased to NOT do that, and instead expects the getter to directly return a OneOfType.Value. If it followed the generic case, I would expect the getter to return a Row here instead. Any thoughts on this? [1] https://github.com/apache/beam/blob/b853f9c7aa3662eb707290ded4c22f86befeda68/sdks/java/core/src/main/java/org/apache/beam/sdk/values/RowWithGetters.java#L140
