Testing Tree Timeline

Hello All,

I was wondering what the current timeline looks like for integrating the patches in the testing tree into the mainline kernel?

Myself, and several other researchers at Ohio University, believe that DCCP would be an ideal protocol for encapsulating Delay Tolerant Networking protocols for transportation across the Internet. For testing purposes, we do not desire guaranteed reliability, but do require congestion control.

After testing (and bug fixing) DCCP for most of the last year, we are convinced that DCCP with CCID2 will be effective for our goals. In fact, we are in the process of writing an Internet Draft to recommend that other researchers use DCCP for encapsulating these DTN protocols.

However, the CCID2 code in the mainline kernel is very buggy and suffers from very poor performance. As a result, the only way to get a version of CCID2 that works well is to build a custom kernel from the testing tree. This, obviously, puts a damper on adoption.

As a result, it would be nice to have some idea of when code typically migrates from the testing tree into the mainline kernel and how that determination is made.

Thanks.

Samuel Jero
Internetworking Research Group
Ohio University

Hi Samuel,

The way kernel code gets upstream is that whoever feels the most need for it will take code and push it upstream. I did this with Arnaldo earlier on and then Gerrit has largely taken over the reign. None of us were doing it as our job but as an interest on the side. This is why it sometimes becomes low priority.

I would encourage you then to try and do the same and send it off to the netdev mailing list with cc to dccp, Gerrit and Dave Miller. Do try to read the patch submission guides, but don't be discouraged when you make mistakes in doing it as you will. You might get "flamed" even but consider that your welcome to the kernel world, fix the mistakes and carry on.

Regards

Ian

Hi Samuel,

you can rest assured that your patches will eventually be submitted to the netdev list and from there to David Miller.

I apologize that this has not progressed faster, but I can only confirm what Ian just said.

There are dependencies between the patches and your set is pretty much at the top, I will need some time to refactor and sort out the dependencies.

I am sorry I can't give you a promise to have all this wrapped up by Monday or so, this is work done out of hours, it may take a few weeks since I like to submit patches in small chunks of about 5-6 in order to respect the workload that the higher-level maintainers have.

So please bear with me, it may just take a little while but we will get there eventually.

Gerrit

Quoting Samuel Jero:

| Hello All,
| 
| I was wondering what the current timeline looks like for integrating the
| patches in the testing tree into the mainline kernel?
| 
| Myself, and several other researchers at Ohio University, believe that
| DCCP would be an ideal protocol for encapsulating Delay Tolerant
| Networking protocols for transportation across the Internet. For testing
| purposes, we do not desire guaranteed reliability, but do require
| congestion control.
| 
| After testing (and bug fixing) DCCP for most of the last year, we are
| convinced that DCCP with CCID2 will be effective for our goals. In fact,
| we are in the process of writing an Internet Draft to recommend that
| other researchers use DCCP for encapsulating these DTN protocols.
| 
| However, the CCID2 code in the mainline kernel is very buggy and suffers
| from very poor performance. As a result, the only way to get a version
| of CCID2 that works well is to build a custom kernel from the testing
| tree. This, obviously, puts a damper on adoption.
| 
| As a result, it would be nice to have some idea of when code typically
| migrates from the testing tree into the mainline kernel and how that
| determination is made.
| 
| Thanks.
| 
| Samuel Jero
| Internetworking Research Group
| Ohio University
|