Because reading a record type requires the decoder to be able to look up oid
to type mapping and types such as hstore have types that are not fixed between
different PostgreSQL servers it was necessary to restructure the pgtype system
so all encoders and decodes take a *ConnInfo that includes oid/name/type
information.
Though this doesn't follow Go naming conventions exactly it makes names more
consistent with PostgreSQL and it is easier to read. For example, TIDOID becomes
TidOid. In addition this is one less breaking change in the move to V3.
When there was an issue with DB server ConnPool.Acquire() used to block
until it heard back from the server or its OpenTimeout hit. If we had
OpenTimeout set to 3 secs, and there were X go routines trying to aquire a
connection in parallel, the very last go routine would receive timeout
error in X*OpenTimeout seconds because of the blocking nature of the
ConnPool.Acquire().
With this commit ConnPool.Acquire() is not blocking any more, and all
X requests will take about 1*OpenTimeout secs to fail.
This is a quick attempt to improve connection startup time by caching
the properties that are loaded when a connection is ready in the pool,
so that further connections don't incur this cost.
I'm not entirely convinced by the interface here, perhaps these 3 items
could live in their own type and that be passed around for clearer code,
but the idea works well.
If no connections are available, and the pool is reset, there will
be no connections at all in the pool. So the pool needs to be able
to create a connection after waiting for a connection to be released.
- Use unlisten * when releasing connection with listeners to pool
- Only unlisten on releasing connection to pool when necessary
- Remove Unlisten("") as was to release all listeners
ErrDeadConn is returned when calling an already dead connection. But the initial failure returns the real error. So we check for IsAlive instead of ErrDeadConn.
Added test for ConnPool.Begin retry logic.
