Fields

• hostname :: Text
   
• port :: Word16
   
• user :: Text
   
• password :: Text
   
• database :: Text
   
• options :: Text
   

Fields

• killedThreadPollIntervalMs :: Int

  How long in ms Hpgsql will sleep before re-checking if active queries have been orphaned from their issuing threads having died. The default is 500ms, and this is only relevant if you plan on concurrently issuing queries on a single connection, and even then only if you expect your threads to be killed by asynchronous exceptions frequently enough, and you want resume using the connection and cannot wait ~500ms until Hpgsql realizes it's fine to do so. You probably don't need to worry about this or tune it.

• cancellationRequestResendIntervalMs :: Int

  How long in ms Hpgsql will wait before re-sending a cancellation request while draining orphaned queries (queries from dead threads). The default is 500ms, and this is only relevant if you plan on interrupting your queries with asynchronous exceptions, either by use of concurrency primitives or functions like timeout, and continue using the connection for new queries after that. It is not recommend setting this below 100ms, because orphaned query draining alternates with resending cancellation requests, so if this is too low it is possible that draining never finishes, leading to a form of livelock.

• fillTypeInfoCache :: Bool

  Immediately after connecting, run a query to fetch all types from the pg_type table. This makes them available in FromPgField instances. The default is True. You should only set it to False if you really know what you're doing, because class instances of custom types can stop working.

Fields

• pgErrorDetails :: Map ErrorDetail ByteString
   
• failedStatement :: !ByteString
   

Fields

• hpgsqlDetails :: Text
   
• innerException :: Maybe SomeException
   
• relatedStatement :: !(Maybe ByteString)
   

Fields

• notifierPid :: !Int32
   
• channelName :: !Text
   
• notifPayload :: !Text
   

Fields

• resetAll :: Bool

  Runs `RESET ALL` and `RESET ROLE` on the connection. Defaults to True.

• unlistenAll :: Bool

  Runs `UNLISTEN *` on the connection and clears the internal queue of notifications. Defaults to True.

• checkTransactionState :: Bool

  Throws an exception if there is an open transaction or if there's a transaction in error state. Defaults to True.

Fields

• typeInfoCache :: TypeInfoCache

  A map with all builtin PostgreSQL types plus user-defined types, unless you specify custom connection options.

Fields

• queryString :: ![SingleQueryFragment]
   
• queryParams :: ![EncodingContext -> (Maybe Oid, BinaryField)]
   
• isPrepared :: !Bool
   

Fields

• queryString :: !ByteString
   
• queryParams :: ![EncodingContext -> (Maybe Oid, BinaryField)]
   
• preparedStmtHash :: !(Maybe String)
   

Fields

• resultColumnTypes :: [(Text, Oid)]
   

Fields

• numRows :: Int64
   

Fields

• rowColumnData :: ByteString
   

Fields

• totalQueriesSent :: !Integer
   
• currentPipeline :: ![QueryState]

  We support only one pipeline sent to the backend at a time.

• mustIssueRollbackBeforeNextCommand :: Bool

  In cases like when an asynchronous exception interrupts a withTransaction section, we want a ROLLBACK to run _before_ any next commands, to preserve the invariant that withTransaction ensures the started transaction is no more regardless of what happens. When such a thing happens, this field is True.

• notificationsReceived :: !(TQueue NotificationResponse)

  How did we end up with a TQueue inside an object that is already a TVar? This is unnecessary and could live out of InternalConnectionState.

• preparedStatementNames :: Set String
   
• transactionStatusBeforeCurrentPipeline :: !TransactionStatus
   

Fields

• queryIdentifier :: !QueryId
   
• queryText :: !ByteString
   
• queryProtocol :: !QueryProtocol
   
• queryPrepStmtName :: !(Maybe String)
   
• queryOwner :: !WeakThreadId
   
• responseMsgsState :: ResponseMsgsReceived

  Storing every single "control" (i.e. not DataRow) message received for each query means we can continue to drain its results from another thread at anytime, which is necessary to continue using the connection in the presence of asynchronous exceptions.

Fields

• socket :: !Socket
   
• socketClosed :: !(MVar Bool)
   
• recvBuffer :: !(MVar ByteString)
   
• sendBuffer :: !(MVar [(ByteString, STM ())])
   
• socketMutex :: !Mutex
   
• originalConnStr :: !ConnectionString
   
• connectedTo :: !AddrInfo
   
• encodingContext :: !(MVar EncodingContext)
   
• parameterStatusMap :: !(MVar (Map Text Text))
   
• internalConnectionState :: !(TVar InternalConnectionState)
   
• connPid :: !Int32
   
• cancelSecretKey :: !Int32
   
• connOpts :: !ConnectOpts