![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
The leak occurs when libpq can not establish a connection to the database server, in my case when it is not running. I believe that when a caller calls PQreset() or PQresetStart(), the subsequent call to connectDBStart() clobbers the previous value of addrlist in the PGconn. Presumably, closePGconn() should be destroying this value. |
#3
| |||
| |||
|
|
"Cade Cairns" <cadec (AT) otii (DOT) com> writes: The leak occurs when libpq can not establish a connection to the database server, in my case when it is not running. I believe that when a caller calls PQreset() or PQresetStart(), the subsequent call to connectDBStart() clobbers the previous value of addrlist in the PGconn. Presumably, closePGconn() should be destroying this value. Uh ... could we see a complete test case for this? regards, tom lane |
#4
| |||
| |||
|
|
I will post a patch in a few days. I am extremely busy and don't have adequate time. I apologize for not having posted more in the first place. On 27-Mar-05, at 1:12 AM, Tom Lane wrote: "Cade Cairns" <cadec (AT) otii (DOT) com> writes: The leak occurs when libpq can not establish a connection to the database server, in my case when it is not running. I believe that when a caller calls PQreset() or PQresetStart(), the subsequent call to connectDBStart() clobbers the previous value of addrlist in the PGconn. Presumably, closePGconn() should be destroying this value. Uh ... could we see a complete test case for this? regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 2: you can get off all lists at once with the unregister command (send "unregister YourEmailAddressHere" to majordomo (AT) postgresql (DOT) org) |
![]() |
| Thread Tools | |
| Display Modes | |
| |