Should the author/committer columns of the Revision / Release table (and other UDTs) be frozen and/or flattened?

With @douardda we've tried issuing a query on the revision table by revision author email.

There are (seemingly) two issues for doing so:

• partial matching of UDTs is not possible (that is, a where clause for a UDT must match all fields exactly)
• non-frozen UDT columns cannot be matched at all (InvalidRequest: Error from server: code=2200 [Invalid query] message="Non-frozen UDT column 'author' (person) cannot be restricted by any relation")

Freezing the UDT columns (which would prevent updates on only one of the fields, which we don't care about at all) would allow querying using them, but would not solve the partial matching issue.

We should consider whether these fields should be frozen (probably yes) and whether they should be flattened into top level table fields (probably yes for the fields that we may want to be able to query on).

added cassandra label

changed title from Should the author/committer columns of the Revision / Release table (and other UDTs) be frozen and/or unrolled? to Should the author/committer columns of the Revision / Release table (and other UDTs) be frozen and/or flattened?

changed the description

I agree with flattening, I don't think there is a good reason to use UDTs except mildly simpler Python code