Refactoring Congestion Control

Hi,
Me and Kentaro had the same idea of refactoring the
congestion  
control algorithms to make them mode modular. (We could, for
 
instance, program TCP/Westwood+ and TCP/Daytona more
easily).
Turns out he already did the work for an old NetBSD version.
I grabbed the patch and updated it to -current.

Any comments before I commit ?

	-- Rui Paulo

On Sun, 17 Sep 2006, Rui Paulo wrote:
> Any comments before I commit ?

Where is the documentation or at least a design document?


  - Hubert

On Sep 17, 2006, at 10:13 PM, Hubert Feyrer wrote:

> On Sun, 17 Sep 2006, Rui Paulo wrote:
>> Any comments before I commit ?
>
> Where is the documentation or at least a design
document?

There's no need for such thing.


	-- Rui Paulo

On Sun, 17 Sep 2006, Rui Paulo wrote:
>> Where is the documentation or at least a design
document?
> There's no need for such thing.

Serious, can you describe what the code does?


  - Hubert

On Sep 17, 2006, at 10:34 PM, Hubert Feyrer wrote:

> On Sun, 17 Sep 2006, Rui Paulo wrote:
>>> Where is the documentation or at least a design
document?
>> There's no need for such thing.
>
> Serious, can you describe what the code does?

Ok. The code is the heart of the congestion control in our
TCP stack.  
I'm moving stuff around and creating a new
"class" a-la objective  
programming (in this case a struct with three function
pointers).
This way we can separate the congestion control algorithms
we have in  
our source tree.
Basically with this we should be able to remove the checks
for if  
(tcp_newreno) or if (tcp_westwood), etc.
Note that I'm not adding new code, hence the subject
"refactoring".

	-- Rui Paulo