| Dart Home | PowerTCP WebServer for ActiveX | Custom Development | Reply | PowerTCP WebServer for ActiveX Topics | Forums |
| Author | Forum: PowerTCP WebServer for ActiveX Topic: I'm really disappointed... |
| User1137 Posts: 6 Member Since: 10/01/00 |
posted October 7, 2000 6:18 PM Why would you take winsock and split it's functionality, REQUIRING a user to use seperate controls for server and client applications? I have an application that I was coding to dial-in, leave a message, disconnect, and start listening for a response due sometime within the next 24 hours from another client. With Crescent's TCP control, this is easy. You just disconnect and set ListenForConnection. But, the only way I have seen to make this work with your controls is to have a Client component dial up and exchange the info, and have a Server component begin listening on a different port, and have the Dart client tcp control pass this Server's listening port to the main server when it leaves it's message. (And, I wouldn't want to disconnect the client, and assign the same port to the server for listening, because it is possible that another process could be assigned that port during the transfer from one control to the other.) That's really messy. Please tell me I am missing something. |
Tony Priest From: Utica, NY USA Posts: 8466 Member Since: 04/11/00 |
posted October 7, 2000 7:42 PM I'm kind of unclear on what you want to do. What do you mean by "pass this Server's listening port to the main server"? What's the "main server"? What's "this server"? What exactly do you mean by "another process could be assigned that port during the transfer from one control to the other"? |
| User1137 Posts: 6 Member Since: 10/01/00 |
posted October 7, 2000 8:39 PM OK... C1 (Computer 1 is an online server). C2 is a client machine, connected to the internet that wishes to interact with C1. First let's cover what I am trying to accomplish. The goal is to have a system in which client machines are notified in near-real-time when an event that they are interested in happens. Constant TCP/IP connections are one way to go, but that's really easy for hackers to clog up, by simply logging on from different ports on the same machine. (Yes, I could restrict logins to one port per IP, but that would cut out people at work, who are mostly behind firewalls or proxy servers.) Constant TCP/IP connections would also require a lot more hardware than just maintaining a list of clients that are listening for their news. And (to help prevent abuse of the system), I want to verify that there is a machine at the IP/Port that logs in, so transferring some type of random data and having it returned is desired. UDP is one option, but the delivery of this data in a relaible fashion is a must. The entire advertising of this venture is based on the idea that you WILL get the news that you have requested as soon as it is reported. Udp is faster, but less reliable. When I send a packet and get no response, how many times do I resend, without getting back acknowledgement from the client? Have they gone offline? Is there a problem between us? When is it enough? C2 should log in to C1, exchanging random data to ensure that some type of client (even if a hacked client) is at the C2 IP/Port address. C1 should record the Port/IP of C2 and a random string for a key. When news that is needed by C2 is found, C1 should use the recorded IP/Port to send the data to C2, including the random string for an identifier. If the random string is wrong, C2 ignores the data. If the random string is correct, C2 requests the appropriate page to view or record the requested news item. Sounds simple enough. The system will be high-profile, so hackers should be considered (hence the random string). Any ideas? I may be able to use a TCP control and a Server control with ReuseAddress set on both and the Server listening on the TCP's port. The documentation for this property is rather weak, so I am not sure if this would work. You'd still have to include a Server control, whereas with Crescent's TCP control you can just set the ListenForConnection property after using it for the first connection and data exchange. (Gee, I really wish yours was that easy.) BTW, love the server control. |
| Tony Priest From: Utica, NY USA Posts: 8466 Member Since: 04/11/00 |
posted October 7, 2000 11:16 PM That doesn't sound all that simple to me. In fact I still don't really understand what you want. Let's take it one issue at a time. Why do you feel that you need to use ReuseAddress? |
| User1137 Posts: 6 Member Since: 10/01/00 |
posted October 7, 2000 11:19 PM Well, I'd like to use the same IP/Port that the Client originally connected to the server on for a TCP listener on the client machine. |
| Tony Priest From: Utica, NY USA Posts: 8466 Member Since: 04/11/00 |
posted October 7, 2000 11:25 PM We're not going to get anywhere on the forum. Try calling in on Monday so that we can discuss this. |
| User1137 Posts: 6 Member Since: 10/01/00 |
posted October 7, 2000 11:32 PM No problem.... Talk to ya Monday. |
| Reply | PowerTCP WebServer for ActiveX Topics | Forums |
This site is powered by
PowerTCP WebServer for ActiveX
|