Yes, well, sorta. UDP is packet based meaning that it is sent in small bundles of data. TCP does this too but it has an extra layer of abstraction to make it streamable. UDPs that are sent will be recv''ed at the same size assuming its smaller than the MTU (or maximum transmission unit) which is usually around 512 but changes. Also note that because of this you might want to avoid sending small fragments of data as sending packets sized 12, 16, 20 and 20 bytes will be slower than sending it all in one datagram.

Or to say it more concretely...

Yes, if you''re using UDP and you send() 100 bytes in a single call then when you do a recv() you''ll get all 100 bytes (and no others) on the other side. In TCP a send() of 100 bytes might be recv()''d one byte at a time, or 5, or all 100, or all 100 plus the first couple of bytes of the next send(), you can''t tell.

You shouldn''t choose UDP or TCP just because of this though. They are very different beasts.

-Mike