I am writing a program in Python that will act as a server and accept data from a client, is it a good idea to impose a hard limit as to the amount of data, if so why?
More info: So certain chat programs limit the amount of text one can send per send (i.e. per time user presses send) so the question comes down to is there a legit reason for this and if yes, what is it?
What is your question exactly?
What happens when you do receive on a socket is that the current available data in the socket buffer is immediately returned. If you give receive (or read, I guess), a huge buffer size, such as 40000, it'll likely never return that much data at once. If you give it a tiny buffer size like 100, then it'll return the 100 bytes it has immediately and still have more available. Either way, you're not imposing a limit on how much data the client is sending you.
Most likely you've seen code which protects against "extra" incoming data. This is often due to the possibility of buffer overruns, where the extra data being copied into memory overruns the pre-allocated array and overwrites executable code with attacker code. Code written in languages like C typically has a lot of length checking to prevent this type of attack. Functions such as gets, and strcpy are replaced with their safer counterparts like fgets and strncpy which have a length argument to prevent buffer overruns.
If you use a dynamic language like Python, your arrays resize so they won't overflow and clobber other memory, but you still have to be careful about sanitizing foreign data.
Chat programs likely limit the size of a message for reasons such as database field size. If 80% of your incoming messages are 40 characters or less, 90% are 60 characters or less, and 98% are 80 characters or less, why make your message text field allow 10k characters per message?
I don't know what your actual application is, however, setting a hard limit on the total amount of data that a client can send could be useful in reducing your exposure to denial of service attacks, e.g. client connects and sends 100MB of data which could load your application unacceptably.
But it really depends on what you application is. Are you after a per line limit or a total per connection limit or what?