From: Warren Young
Date: March 7 2008 7:43am
Subject: Re: Strange behavior of connect() when specifying a different port
List-Archive: http://lists.mysql.com/plusplus/7502
Message-Id: <47D0F23F.7000802@etr-usa.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit

António Santos wrote:
> 
> Amazingly, no matter what port number I use, it always connect! 

Then it isn't using TCP/IP, which is the only way to get it to use a 
port number.  You can verify this with netstat.  While the program is 
connected and running:

	$ netstat -na |grep 3306

You should see nothing ESTABLISHED while it's connected with an invalid 
port argument.  (Assuming you have complete control of the machine so 
you can ensure all other database users are stopped while testing.)

I did some experimenting and found that the C API uses Unix domain 
sockets (mysql.sock) if you use 'localhost'.  If you use 127.0.0.1, it 
uses TCP/IP, here at least.  You could also use the machine's "real" IP 
address to be sure, at the possible cost of some inefficiency, depending 
on how the kernel routes the calls.  If it has to bounce through the 
Ethernet driver, it's better to use 127.0.0.1 if you can.

> I installed mysql++ using yum.

Try MySQL++ v3.0.  I just said this here:

	$ ./exrun resetdb -uroot -pupward1 -s127.0.0.1:3307

and got "Connection failed: Can't connect to MySQL server on '127.0.0.1' 
(111)"  This is on a CentOS 5.1 system running the official MySQL 5.0.45 
binaries.  If I leave the :3307 off, change it to 3306, or use 
'localhost', it does connect.