Re: SOCKET Connect() connects but shouldn't. -- RPG400-L


On 04/02/2006, at 1:57 PM, Dave Johnson wrote:

I am trying to figure out some basic sockets programming using ScottKlement's tutorial. This is an excellent tutorial!
Here is the problem I am finding.

I use the code from Scott's sample:


Code deleted.

and it works fine. The problem is that it shouldn't work because I donot actually have a socket server listening on the port I specify inthe SIN_PORT paramater. That is, if I use a port # of 8540 in SIN_PORTfor example but my socket server is actually listening on port 3994,then I should get an error, right? But I don't. Instead, I get thereturn value = 0 and the program continues on.
If it makes a difference, my socket server is running on a pc and Iuse the DOS command NETSTAT -a to see what is active on all the ports.I see my socket server listening on port 3994 and nothing on 8540,and yet the connect does not return a negative value.
Obviously I am missing something here, but I am so new at this stuffthat I cannot see what it is.
Anyone have any ideas of things I could check?


Couple of things:

First, byte-ordering. The bytes for network values must be in networkbyte order. This happens to be big-endian but PCs and other toys uselittle-endian thus your port number is probably being misinterpreted.You should always use the byte-order conversion functions when settingvalues in a socket address structure.


         c                   eval      sin_addr = htonl(IP)
         c                   eval      sin_port = htons(port)

The example code you showed will work on OS/400 because it is abig-endian system.

Given your specified port of 8540 which in hex is 215C which inlittle-endian is 5C21 which will be interpreted as port 23585 by thesocket code on a PC. A similar misinterpretation will occur with the IPaddress so you may be attempting to connect to a completely differentserver which does happen to have a listener on port 23585.

Secondly, it is not apparent from your code but I presume IP is a32-bit integer value and NOT the dotted-decimal form of the address. Ifit is the dotted-decimal address (i.e., a character value) then youwill have to convert from the dotted-decimal form to the network formusing inet_addr or inet_aton (or inet_pton if you want IPv6compatibility).

Thirdly, you should be checking errno when a socket function indicatesan error condition. This will give you a better** indication of thereason for the failure. For example, ETIMEDOUT, EHOSTUNREACH, orENETUNREACH if no server is found or ECONNREFUSED if no listener isactive on the target port.


Try running the example code on OS/400 and see what happens.

**better is a relative term: in this case it means better by Unixstandards which are quite pathetic by OS/400 standards but better thannothing.


Regards,
Simon Coulter.
--------------------------------------------------------------------
   FlyByNight Software         AS/400 Technical Specialists

   http://www.flybynight.com.au/
   Phone: +61 3 9419 0175   Mobile: +61 0411 091 400        /"\
   Fax:   +61 3 9419 0175                                   \ /
                                                             X
                 ASCII Ribbon campaign against HTML E-Mail  / \
--------------------------------------------------------------------