replacing functional decomposition design with object oriented

[mrbean]

So, based on your link to that other post, I should try and use ports above 49152 to ensure anonimity.  Any thoughts on my OO question about the RCTx becoming a Factory (design pattern) of sorts.

[mrbean]

As I've gone through this, I've started to question why I need more than one TCP connection.  I mean, in theory, since (in the Multiple Connections example) all clients us the same localhost::port, it's really the same connection.  In my application, the client is the same VI - it just needs to talk to possibly sixteen radios.  So instead of opening a new TCP connection, I could probably just pass a unique identifier (to specify which radio), and the RCTx would then know which radio to which it has to forward the message.  Where things would get messy, is in the RCTx.  It will have to create multiple instances of the radio interface so as to pass the client message to the right destination.   I'm not sure if I want to try to tackle this with a monster cluster/array or begin an OO approach.

[Mark_Yedinak]

---

mrbean wrote:
So, based on your link to that other post, I should try and use ports above 49152 to ensure anonimity.  Any thoughts on my OO question about the RCTx becoming a Factory (design pattern) of sorts.

---

Technically speaking, yes you should use port numbers above 49152. In times past in was generally said to use port number above 1024. Anything below that and you will most likely run into problems. Most of teh reserved ports above 1024 are highly specialized and you will generally not find them on most systems. But to be safe it would be best to follow the IETF guidelines.