August 23rd, 2003, 01:45 AM
Whoa, man, that's amazing. It works perfectly - it's written in c++, but compiled in Lcc without errors...
It is NOT written in c++. That's why compiled so good :)
Last edited by Loser; August 31st, 2003 at 10:20 AM.
August 25th, 2003, 02:30 AM
Just another question:
Is there a way to determine which IP address from the list is "outside" - to address a computer over the internet?
August 25th, 2003, 11:15 AM
Infamous has started a thread addressing this problem, in which he's looking at a variant of the traceroute utility. I'm not sure how far he's gotten yet.
August 25th, 2003, 11:21 AM
Huh, it's not so important. If this is so complicated, I'll just write all addresses and then the user will have to try a little. In the end, he's not as stupid as I am. :)
August 31st, 2003, 10:18 AM
What's the most optimal "packet" (string) size (length) to send over tcp?
August 31st, 2003, 10:26 AM
Are you trying to build your own packets or just send data? If you are sending data from an application, just send as much as you have at any given time and the underlying levels will break them into acceptable sizes. If you are trying to build your own packets (why?), then you still don't have a lot of control because the maximum packet size depends on the route betwixt you and your destination. Unless you set the 'do not fragment' flag in your packet, it will be broken into the smallest size that can be accomodated by the route your packets happen to take (which can vary, even during a single transaction).
The easy answer is: What ever size your buffer is (keeping in mind that bigger is better, it is a dumb idea to send it one byte at a time, though the lower layers may wait a bit to accumulate data).
August 31st, 2003, 10:37 AM
Don't worry, I won't build my own packets. :)
So 16384 bytes is OK?
Thanks for the answer.
August 31st, 2003, 11:28 AM
I use exactly the same buffer size. Sometimes, when testing a client/server pair on the same computer, it even transmits and recieves that number of bytes. I imagine (though haven't looked) that the actual packet size is much smaller (if I recall, the stated maximum TCP packet size is 1,500 bytes). Still, since the underlying socket implementation is supposed to take care of all that for you, it doesn't make you no nevermind.
The idea is to give the socket layer data in chunks at least the size of a packet so that it doesn't send a bunch of tiny packets (it takes almost as long to send one byte as a full sized packet).
August 31st, 2003, 01:27 PM
I think this topic should end now. :)
Again, thanks for the answers.