I was thinking about what algorithms to use on a network I’m programming. The network is quite simple (serial port / modem). I’ve been trying to research some of this stuff on Google, but for some reason Google is not responding, and hasn’t been for the last week. I’m not asking you to research for me, just give me a little info on something similar you may have done, or a link that might help me. I’m thinking about creating threads for input and output. The threads use linked lists for data packets. The data is passed through the network and it is appended to the top of the stack. It is client program’s job to read and use this information in the list. The packet list will not be accessed directly, however, indirectly. What I mean is, you can’t access the nodes with in the list, but a point to a copy of a node will be returned instead. I’m thinking about adding some popping and pushing methods for each query or something. As you may see, my design needs work. It’s not a design really, I’m just trying to organize the information I’m learning and practice using it. I’ll start small and work up into TCP/IP networks. Until then, please give me your info.

TCP/IP is like a stream. You can read the data in whatever sized chunks you want; regardless of what sized chunks you sent it in.

If you send "Hello", then " ", and then "World!", the other computer can read it all at once as one chunk, i.e., "Hello World!", or a character at a time, i.e., "H", "e", "l", "l", "o". . . or whatever.

Anyway, the point I'm trying to make is that storing data you read from TCP/IP as packets doesn't make any sense.

Now; UDP/IP on the other hand; then packets would make sense.

Google is just fine, I use it 10-20 times a day. Make sure you don't have any spyware installed which redirects away from it.