Re: [Ecn-sane] please commit anything new you got

Re: [Ecn-sane] please commit anything new you got

[Jonathan Morton]

> On 17 Mar, 2019, at 12:04 am, Jonathan Morton <chromatix99@gmail.com> wrote:
> 
> I added something about single-queue AQMs which is definitely food for thought.

…and now I've propagated those thoughts more widely in the relevant documents.  Now would be a great time to look at the Git history, and start thinking about how much work it'd take to turn out a couple more I-Ds.

I've even added brief notes on how to implement DCTCP-SCE.  It's very easy if you have a working feedback path and a clean NewReno to start with.  As for the feedback path, I looked into whether AccECN could take more optimisation for SCE than just rearranging the fields, but since it completely replaces the relevant TCP flags, it has to rely entirely on the TCP option.

 - Jonathan Morton

[Ecn-sane] copyright notice on sce.h?

[Dave Taht]

We've got the primary authors added to sce.h, which I haven't
committed yet. (jonathan, me, pete, dave reed). Would
anyone else here like in? (please email me privately). I always like
to give credit to all the sources of a good idea....

The authors of sce.h for bsd are logan and rodney, presently.

Once we have that settled we can submit to mainline, add it to
sch_cake, etc, etc.

On Wed, Mar 20, 2019 at 11:48 AM Jonathan Morton <chromatix99@gmail.com> wrote:
>
> > On 17 Mar, 2019, at 12:04 am, Jonathan Morton <chromatix99@gmail.com> wrote:
> >
> > I added something about single-queue AQMs which is definitely food for thought.
>
> …and now I've propagated those thoughts more widely in the relevant documents.  Now would be a great time to look at the Git history, and start thinking about how much work it'd take to turn out a couple more I-Ds.
>
> I've even added brief notes on how to implement DCTCP-SCE.  It's very easy if you have a working feedback path and a clean NewReno to start with.  As for the feedback path, I looked into whether AccECN could take more optimisation for SCE than just rearranging the fields, but since it completely replaces the relevant TCP flags, it has to rely entirely on the TCP option.
>
>  - Jonathan Morton
>


-- 

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740

Re: [Ecn-sane] copyright notice on sce.h?

[Rodney W. Grimes]

[ Charset UTF-8 unsupported, converting... ]
> We've got the primary authors added to sce.h, which I haven't
> committed yet. (jonathan, me, pete, dave reed). Would
> anyone else here like in? (please email me privately). I always like
> to give credit to all the sources of a good idea....
> 
> The authors of sce.h for bsd are logan and rodney, presently.

Logan only, I wont take credit for what I may do to put it
into a more proper form as for me it is just a mechancial
thing, aka treat me more as an editor.

> Once we have that settled we can submit to mainline, add it to
> sch_cake, etc, etc.
> 
> On Wed, Mar 20, 2019 at 11:48 AM Jonathan Morton <chromatix99@gmail.com> wrote:
> >
> > > On 17 Mar, 2019, at 12:04 am, Jonathan Morton <chromatix99@gmail.com> wrote:
> > >
> > > I added something about single-queue AQMs which is definitely food for thought.
> >
> > ?and now I've propagated those thoughts more widely in the relevant documents.  Now would be a great time to look at the Git history, and start thinking about how much work it'd take to turn out a couple more I-Ds.
> >
> > I've even added brief notes on how to implement DCTCP-SCE.  It's very easy if you have a working feedback path and a clean NewReno to start with.  As for the feedback path, I looked into whether AccECN could take more optimisation for SCE than just rearranging the fields, but since it completely replaces the relevant TCP flags, it has to rely entirely on the TCP option.
> >
> >  - Jonathan Morton
> >
> 
> 
> -- 
> 
> Dave T?ht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-205-9740
> _______________________________________________
> Ecn-sane mailing list
> Ecn-sane@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/ecn-sane
> 
> 

-- 
Rod Grimes                                                 rgrimes@freebsd.org

Re: [Ecn-sane] copyright notice on sce.h?

[Dave Taht]

I have committed a reviewed version of sce.h to the
https://github.com/dtaht/fq_codel_fast

repository. Please retest.

On Wed, Mar 20, 2019 at 11:02 PM Rodney W. Grimes
<4bone@gndrsh.dnsmgr.net> wrote:
>
> [ Charset UTF-8 unsupported, converting... ]
> > We've got the primary authors added to sce.h, which I haven't
> > committed yet. (jonathan, me, pete, dave reed). Would
> > anyone else here like in? (please email me privately). I always like
> > to give credit to all the sources of a good idea....
> >
> > The authors of sce.h for bsd are logan and rodney, presently.
>
> Logan only, I wont take credit for what I may do to put it
> into a more proper form as for me it is just a mechancial
> thing, aka treat me more as an editor.
>
> > Once we have that settled we can submit to mainline, add it to
> > sch_cake, etc, etc.
> >
> > On Wed, Mar 20, 2019 at 11:48 AM Jonathan Morton <chromatix99@gmail.com> wrote:
> > >
> > > > On 17 Mar, 2019, at 12:04 am, Jonathan Morton <chromatix99@gmail.com> wrote:
> > > >
> > > > I added something about single-queue AQMs which is definitely food for thought.
> > >
> > > ?and now I've propagated those thoughts more widely in the relevant documents.  Now would be a great time to look at the Git history, and start thinking about how much work it'd take to turn out a couple more I-Ds.
> > >
> > > I've even added brief notes on how to implement DCTCP-SCE.  It's very easy if you have a working feedback path and a clean NewReno to start with.  As for the feedback path, I looked into whether AccECN could take more optimisation for SCE than just rearranging the fields, but since it completely replaces the relevant TCP flags, it has to rely entirely on the TCP option.
> > >
> > >  - Jonathan Morton
> > >
> >
> >
> > --
> >
> > Dave T?ht
> > CTO, TekLibre, LLC
> > http://www.teklibre.com
> > Tel: 1-831-205-9740
> > _______________________________________________
> > Ecn-sane mailing list
> > Ecn-sane@lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/ecn-sane
> >
> >
>
> --
> Rod Grimes                                                 rgrimes@freebsd.org



-- 

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740