VoIP Mailing List Archives

[nik.middleton at noble...]

Hi Guys,

I’m baffled by this error. I’m updating the db on call hang-up If I comment out curs:close() no error, but I’m concerned about memory leaks. Can anyone tell me what FS is complaining about?

The db gets updated in both cases

Regards



require "luasql.mysql"

function myHangupHook(s, status, arg)
freeswitch.consoleLog("info", " : They hung up on US!!!\n");
env = assert (luasql.mysql())
con = assert (env:connect("xxxxl","xxxxxxxxx","pass","192.168.3.205"))
curs = assert (con:execute"UPDATE callers SET lastcall = 'BOB' WHERE id = 33292")
curs:close()
con:close()
env:close()
freeswitch.consoleLog("NOTICE", "myHangupHook: " .. status .. "\n");
--error()
end




2009-02-10 20:53:20 [INFO] switch_cpp.cpp:1086 console_log() : They hung up on US!!!
2009-02-10 20:53:20 [ERR] mod_lua.cpp:176 lua_parse_and_execute() /usr/local/freeswitch/scripts/helloworld.lua:50: attempt to index global 'curs' (a number value)
stack traceback:
/usr/local/freeswitch/scripts/helloworld.lua:50: in function </usr/local/freeswitch/scripts/helloworld.lua:45>
[C]: in function 'hangup'
/usr/local/freeswitch/scripts/helloworld.lua:70: in main chunk
2009-02-10 20:53:20 [NOTICE] switch_core_session.c:957 switch_core_session_thread() Session 63 (sofia/internal/1001@192.168.3.206) Ended

[c_cav_01 at yahoo.com]

Closing the connection will force the server to close any open transactions, as well as release recordsets in local memory that reference the connection.

However curs is not a recordset. An SQL update is going to return an integer (rows affected) or boolean depending on the which server you use since no recordset is actually requested.

--- On Tue, 2/10/09, Nik Middleton <nik.middleton@noblesolutions.co.uk> wrote:

[nik.middleton at noble...]

So what you’re saying is that I can comment out curs:close() as it’s not needed?

Regards,




From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Chris
Sent: 10 February 2009 21:19
To: freeswitch-users@lists.freeswitch.org
Subject: Re: [Freeswitch-users] Strange error message

Closing the connection will force the server to close any open transactions, as well as release recordsets in local memory that reference the connection.



However curs is not a recordset. An SQL update is going to return an integer (rows affected) or boolean depending on the which server you use since no recordset is actually requested.

--- On Tue, 2/10/09, Nik Middleton <nik.middleton@noblesolutions.co.uk> wrote:

[c_cav_01 at yahoo.com]

In that particular instance, no, it's not needed. It's assigned a numeric value and is just a variable that will disappear when it goes out of scope. You could deallocate or release the variable if you wished to be tidy.

In the case of a recordset being returned by the server (an SQL "SELECT") then it will be a recordset object and would need to be closed.

If you look at your error, your trying to index a number. e.g curs assigned numeric by the UPDATE returning the number of rows affected, then your trying to look for a :close method or property on a numeric var which isn't going to work and the compiler is interpreting that as an attempt to index it.

In general, on most SQL servers, SQL insert, update and delete calls return numerics, only select returns recordsets.

--- On Tue, 2/10/09, Nik Middleton <nik.middleton@noblesolutions.co.uk> wrote: