(sql:pq-finish connection)

PQfinish the connection and mark it as invalid

(sql:pq-clear result)

PQclear the result and mark it as invalid

(sql:sql-error connection result format-string &REST arguments)

finalize connection and result and SIGNAL an appropriate ERROR

(sql:sql-connect &KEY host port options tty name login password)

call PQsetdbLogin and return the connection

(sql:with-sql-connection (variable &REST options &KEY log &ALLOW-OTHER-KEYS) &BODY body)

1. bind *sql-log* to the log argument

2. call sql:sql-connect on options and bind variable to the result

3. execute body

4. call sql:pq-finish on variable

(sql:sql-transaction connection command status &OPTIONAL (clear-p T))

execute the command via connection; if the status does not match status, ERROR is SIGNALed; if clear-p is non-NIL sql:pq-clear the result; otherwise return it

(sql:with-sql-transaction (result connection command status) &BODY body)

execure the body on the result of command, then sql:pq-clear the result

sql:*sql-login*

the default login argument to sql:sql-connect (initially set to "postgres")

sql:*sql-password*

the default password argument to sql:sql-connect (initially set to "postgres")

sql:*sql-log*

when non-NIL, should be a STREAM; sql:sql-connect and sql:sql-transaction will write to it (initially set to NIL)

Warning

Since PQfinish and PQclear cannot be called on the same pointer twice, one needs to track their validity (sql:sql-connect and sql:sql-transaction take care of that). See Example 32.11, "Controlling validity of resources".