Avro and Parquet and Arrow are almost loseless - they can't handle timestamp with timezone which really sucks for processing world-wide data. I'm still looking

Avro and Parquet and Arrow are almost loseless - they can't handle timestamp with timezone which really sucks for processing world-wide data. I'm still looking for somwthing that can do it. (Of course, Spark doesn't support timestampz which is probably why the formats don't.)