Joe Conway

This library does not implement the cleartext password authentication method in the frontend/backend protocol; it only implements the md5 password method. If you wanted to implement this method, check out...

Planned - no, not really. We'll eventually get to every Postgres type, but priority is based on features needed to support other projects. I like the syntax you have proposed...

It looks more daunting than it actually is: https://www.postgresql.org/docs/9.1/static/protocol.html. Those steps should support both reading and writing. The regex for type may have to be a more inclusive on characters...

I'm not sure what this issue is in reference to. Prepared statements and parameterized queries are already implemented; this occurs in the backend/frontend protocol communication. I'm not sure if this...

Sure, but that is intentional. This issue sounds more a documentation issue?

It is also possible to publicize this information (since it already exists) through another mechanism. I believe the cache for statement reuse also contains this information once a query has...

👍 Makes sense - each row could have a reference to the metadata. Row could implement `List` as well to make the change non-breaking.

I thought it was implemented? Here are supported types AFAIK: https://github.com/stablekernel/postgresql-dart/blob/eebfc7387d3e2af088efb01fb7248cb9e6ca1fe8/lib/src/binary_codec.dart#L283

Did you suffix your substitution variable? e.g. `INSERT INTO t (v) VALUES (@v:bytea)` That's the only way the driver knows to use the binary encoder IIRC.

Looks like `execute` will always use the text protocol and `query` will use the binary protocol. I don't recall why that is or if it should remain that way.