select help

Hi.
How can i find out when a client closed a socket, using select.

I believe (not at home, so I can't confirm) that you place the socket you want to check for being closed in the set to be checked for reading. (As if you wanted to check if data had arrived on the socket.) Then, in your call to recv(), if the socket has been closed, (and after you've recv()'d any waiting data), it'll return 0 for closed. (And maybe <0 on error, but I'm not sure.)

See the MSDN page on select() as well as the unix man page. Shouldn't be too hard to whip up an example yourself to test.

Quote:

---

Originally Posted by Cactus_Hugger

I believe (not at home, so I can't confirm) that you place the socket you want to check for being closed in the set to be checked for reading. (As if you wanted to check if data had arrived on the socket.) Then, in your call to recv(), if the socket has been closed, (and after you've recv()'d any waiting data), it'll return 0 for closed. (And maybe <0 on error, but I'm not sure.)

See the MSDN page on select() as well as the unix man page. Shouldn't be too hard to whip up an example yourself to test.

---

Thanks.
By the way, another question.
In a TCP session, i want that the server make 2 consecutive writes, how can i read this in the client.
If i make 2 read's the second answer is never received by the client.

I don't think I can give you an answer with that... we'd need to see some code. Generally, the method you describe works. Common errors including things like not understanding how send()/recv() work, mainly that they are not guarenteed to actually send()/recv() the amount you tell them to. (See the man pages)