Dynamic buffer size in socket communication - good or bad idea?

Question

is it a good or a bad idea to use dynamic buffer sizes in socket communication?

for example: i have a message format [2 bytes: length of the following message][message]

So the buffer size will be message.length+2. On the receiver side I can catch the first two bytes and then I know how many more i will have to to catch.

Is this a good or a bad idea - and WHY?

Answer 1

It's just unnecessary. There are read() methods where you define the offset into the buffer and the maximum read length, and there are plenty of other places where you can use a byte[] array, an offset, and a length in the API as well. You don't need to be allocating buffers of exactly the right size.

Answer 2

First, its always better to come up with the data structures, that will be used for communication over sockets. Where u define the place holders for message type, size of the message and the actual message.

Before Writing these kind of structures, please keep following in mind:

1. Size utilization.

2. Is cross platform supported, to avoid endianess problems.

3. Is this extendable.

Answer 3

Its not bad or good but its matter of choice whether to use the way you are using for separating out messages in socket-programming, Or use a delimiter Or use a fixed sized message. At times its use case dependent.

Here you can read more