Hi all,

We currently parse prepare one configuration for each connection. This 
configuration contains all the information needed to setup the transport as 
well as the driver itself.
The thing I see with this, is that the driver needs to know about all 
transports. Now imagine we come up with some “quantum-entanglement” transport … 
we would need to extend all drivers that theoretically support that.
I never really was quite happy with this, I have to admit, but it worked for 
now.

When building PLC4C I implemented things differently.

Here I have a parser that goes through the properties in the connection url and 
makes this information available.
Now I get the driver depending on the url and this creates and initializes a 
driver-config structure … now this can update the default properties in there 
with overrides from the connection-url-parameter.
After that’s done the driver knows which transport he needs and accesses that. 
This creates it’s own config structure and initializes that and then overrides 
from the connection url can be applied.

What I’m saying is: Currently we have one config … I would like to split this 
up into a “ConnectionConfig” and a “TransportConfig” … this should really 
separate things cleanly.

What do you think?

Chris

Reply via email to