Re: [Cake] Cake on LEDE - package maintainer required

Re: [Cake] Cake on LEDE - package maintainer required

[Georgios Amanakis]

Dear Dave,

I am one of these happy users of cake (particularly of the ingress mode
added to cobalt), and I would very much like to see cake pushed into
the linux kernel. And I would be more than happy to throw in at least
for the T-shirt. For the time being, I am maintaining cake for
archlinux in the AUR repository.

Best regards,
George

[Cake] Cake on LEDE - package maintainer required

[Kevin Darbyshire-Bryant]

Hi All,

This email finds me looking for a volunteer/s to take over the ownership 
of the cake qdisc module package in LEDE.

It's not particularly onerous, in essence making sure the package points 
at the git commit considered 'stable'.

In theory there's a 'matching' user space patch for 'tc' in the 
'iproute2' package to ensure it agrees with the kernel space qdisc.

My advice is for everyone to take a step back very quickly...as the last 
person left forward of the line will get it ;-)

Cheers,

Kevin

Re: [Cake] Cake on LEDE - package maintainer required

[Stephen Hemminger]

On Tue, 13 Jun 2017 18:41:05 +0100
Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> wrote:

> Hi All,
> 
> This email finds me looking for a volunteer/s to take over the ownership 
> of the cake qdisc module package in LEDE.
> 
> It's not particularly onerous, in essence making sure the package points 
> at the git commit considered 'stable'.
> 
> In theory there's a 'matching' user space patch for 'tc' in the 
> 'iproute2' package to ensure it agrees with the kernel space qdisc.
> 
> My advice is for everyone to take a step back very quickly...as the last 
> person left forward of the line will get it ;-)
> 
> Cheers,
> 
> Kevin

I will keeping asking the questions.
If it works and has value, why is it not upstream?
If is not upstream, does it still matter?

Re: [Cake] Cake on LEDE - package maintainer required

[Kevin Darbyshire-Bryant]

On 13/06/2017 23:02, Stephen Hemminger wrote:
> On Tue, 13 Jun 2017 18:41:05 +0100
> Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> wrote:
> 
>> Hi All,
>>
>> This email finds me looking for a volunteer/s to take over the ownership 
>> of the cake qdisc module package in LEDE.
>>
>> It's not particularly onerous, in essence making sure the package points 
>> at the git commit considered 'stable'.
>>
>> In theory there's a 'matching' user space patch for 'tc' in the 
>> 'iproute2' package to ensure it agrees with the kernel space qdisc.
>>
>> My advice is for everyone to take a step back very quickly...as the last 
>> person left forward of the line will get it ;-)
>>
>> Cheers,
>>
>> Kevin

Not sure if this is aimed at me specifically, however I shall give a
personal, from the heart and from my perspective, full of my opinion, reply.

> 
> I will keeping asking the questions.

They're good questions.

> If it works and has value, why is it not upstream?

It works and has value to me, specifically in the NAT aware per internal
host fairness on both ingress & egress.  It is not upstream because the
person who really understands how it works hasn't pushed it upstream.

If I were in that person's position, I know I'd not be looking forward
to the process of trying to do so, but then I'm somewhat 'burnt out'
from similar processes of trying to get some sensible improvements into
some user space code.

> If is not upstream, does it still matter?

To me, yes, because it's the only fair queueing AQM of which I'm aware
that'll have a go at per-IP fairness all in one easily configurable box.

KDB

Re: [Cake] Cake on LEDE - package maintainer required

[Dave Taht]

I have long sought sufficient funding to push cake into the mainstream
and come up empty. Cake represents thousands of hours of volunteer labor
as it is, and everyone that has worked on it deserves kudos...

... and a check...

or at least... a bloody t-shirt. And I mean that both in the perjorative
sense and in the descriptive sense, of being covered in scars.

Honestly, I'd settle for some happy set of users of cake to throw in for
the t-shirt, at the very least.

Stephen Hemminger <stephen@networkplumber.org> writes:

> On Tue, 13 Jun 2017 18:41:05 +0100
> Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> wrote:
>
>> Hi All,
>> 
>> This email finds me looking for a volunteer/s to take over the ownership 
>> of the cake qdisc module package in LEDE.
>> 
>> It's not particularly onerous, in essence making sure the package points 
>> at the git commit considered 'stable'.
>> 
>> In theory there's a 'matching' user space patch for 'tc' in the 
>> 'iproute2' package to ensure it agrees with the kernel space qdisc.
>> 
>> My advice is for everyone to take a step back very quickly...as the last 
>> person left forward of the line will get it ;-)
>> 
>> Cheers,
>> 
>> Kevin
>
> I will keeping asking the questions.
> If it works and has value, why is it not upstream?
> If is not upstream, does it still matter?

It's pretty much the default for me 'round here. It solved all the
remaing problems the sqm-scripts had. 

> _______________________________________________
> Cake mailing list
> Cake@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cake

Re: [Cake] Cake on LEDE - package maintainer required

[Stephen Hemminger]

On Wed, 21 Jun 2017 17:42:20 -0700
Dave Taht <dave@taht.net> wrote:

> I have long sought sufficient funding to push cake into the mainstream
> and come up empty. Cake represents thousands of hours of volunteer labor
> as it is, and everyone that has worked on it deserves kudos...
> 
> ... and a check...
> 
> or at least... a bloody t-shirt. And I mean that both in the perjorative
> sense and in the descriptive sense, of being covered in scars.
> 
> Honestly, I'd settle for some happy set of users of cake to throw in for
> the t-shirt, at the very least.
> 

Open Source developers need a day job. Expecting a direct check for hard work in
new technology never seems to turn out well. That seems to be one of the root
causes of the GRsecurity complaints.

Last I checked the issues were not that Cake needed more technical tweaks, more
that it needed to changed into upstream kernel style. Should be semi automatic.
Also getting simplified documentation and man pages.

There is a difference between good enough and complete. Software is never
complete.

The other change of pace would accepting the upstream version as definitive
and eliminating forks.

Re: [Cake] Cake on LEDE - package maintainer required

[Outback Dingo]

Ill take it....

On Tue, Jun 13, 2017 at 1:41 PM, Kevin Darbyshire-Bryant
<kevin@darbyshire-bryant.me.uk> wrote:
> Hi All,
>
> This email finds me looking for a volunteer/s to take over the ownership of
> the cake qdisc module package in LEDE.
>
> It's not particularly onerous, in essence making sure the package points at
> the git commit considered 'stable'.
>
> In theory there's a 'matching' user space patch for 'tc' in the 'iproute2'
> package to ensure it agrees with the kernel space qdisc.
>
> My advice is for everyone to take a step back very quickly...as the last
> person left forward of the line will get it ;-)
>
> Cheers,
>
> Kevin
> _______________________________________________
> Cake mailing list
> Cake@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cake

Re: [Cake] Cake on LEDE - package maintainer required

[Kevin Darbyshire-Bryant]

On 28/06/17 00:58, Outback Dingo wrote:
> Ill take it....
> 

Thank you, appreciated.

I can update my patch into LEDE where I drop maintainer to change to 
you, or you can generate your own patch to take over.... up to you.

The PKG_MAINTAINER field does require a real name though.

Cheers,

Kevin