Re: [PATCH - test-tree] DCCP: Increment sequence number on retransmitted the first REQUEST

[Gerrit Renker <gerrit@xxxxxxxxxxxxxx>]

| One thing to always have in mind is that we should, as long as it
| remains sane, to try to have the same code structure as TCP, so that
| people that are used to the TCP code can read the DCCP more easily and
| to identify areas that are are common.
| 
| So when trying to change things like this check if TCP has a similar
| function and if the change would make sense there too.
|  
More than happy to do that since the TCP code is so very good. 

I was wondering if it would be possible to convert part of CCID-2 to directly call
TCP functions? 

There is a lot of overlap, and much in the CCID-2 specification is directly from TCP.

So far there are at least three cases where CCID-2 has benefited from TCP:

 * the computation of the RTO (with the clamping to 200ms), this is already in the test tree
   http://eden-feed.erg.abdn.ac.uk/cgi-bin/gitweb.cgi?p=dccp_exp.git;a=commit;h=2362d489189bf73b47ff778f7f5eca1daaea5f7b

 * Congestion Window Validation (CWV, RFC 2861)
   This is still in my development tree, as I have not had time to write up an experimental
   evaluation. I have had good experiences on a wireless link with it (where the Acks arrive
   in a bundle since the communication is not full duplex); cwnd growth was really well-behaved.

 * have submitted one patch for RFC 3390 conversion where CCID-2/TCP could use the same function:
   http://eden-feed.erg.abdn.ac.uk/cgi-bin/gitweb.cgi?p=dccp_exp.git;a=commitdiff;h=22b6d875e14b70e3959641512c15cccef4f90692

With regard to the DCCP retransmissions, will reply in other thread.
Gerrit   
--
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html