Fwd: Proper query implementation for Postgresql driver

The idea of using extended query protocol for non-prepared queries raised another "best practices" question, which I'm hoping you can help with. To get rid of text encoding (which is locale-dependent, inefficient etc.) for certain fields, it seems that we have to get rid of it for *all* fields. This is because we send queries without knowing their result columns in advance, and would like to pack all messages into a single packet to avoid roundtrips. In other words, when we send a Bind message we don't yet have a RowDescription, and so we can' specify on an individual columnar basis which fields should be returned as binary and which as text. We're left with the option of either going "full binary" or having to support text encoding for *all* fields - the abolition of which was the whole point.

We can split the query process into two roundtrips - wait for the RowDescription and only then send Bind - in order to pick-and-choose text/binary on a columnar basis, but this is a non-trivial performance hit which we want to avoid.

Regarding the option of going "full binary"; selecting all types from pg_type shows hundreds of types. We'd have to provide binary encode/decode implementation for most (?) of them. In addition, say a new type is added (via an extension for example); text encoding at least had the advantage of not forcing us to support everything: the unknown type would be transferred as text and we'd provide that text to the user as an unprocessed string. Going full binary seems to eliminate this possibility.