CyberSamurai0
commented
1 day ago We should also decide an MTU for our packets. As I noted in #12, I think sticking with the TCP standard 1500-byte MTU might be a good idea. Our sample data is not likely to ever reach this data, but a maximum is worth deciding.
The downside is if we use a 1500-character buffer, that's a lot of memory spent each clock cycle. It's important to note that at an average of ~6 chars per word, this gives us a maximum 250 words per Packet. That's far more than we need. Additionally, messages sent in long chunks are significantly more likely to require retransmission... Maybe we should just keep it small and implement fragmentation?
Design and document a standard structure for transmitted data utilizing even parity bit checking and UART-standard start/stop bits.
POTENTIALLY Design and document a means of implementing "packet headers":
It may be worthwhile to stick to a UDP model which does not have any retransmission mechanism. However, due to the unreliability of our medium, it is likely worth enabling retransmission so that the demonstration goes well.