[Vote] Re-open staging for contributions at RC3

[Vote] Re-open staging for contributions at RC3

[Lars Kurth]

[-- Attachment #1.1: Type: text/plain, Size: 2156 bytes --]

Hi all,

I wanted to kick off a vote related to the following threads
* http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html <http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html> - [URGENT RFC] Branching and reopening -unstable
* http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00543.html <http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00543.html> - [xen 4.6 retrospective] [urgent] rename "freeze" window and make release branch as soon as possible after RC1

Although there was no consensus in the general case to say we should always branch at RC-x at any given release, there seems to be enough consensus for branching earlier, given a number of conditions are met:

In particular:
1: We should not re-open staging too early (aka we would need to get a sense how much churn to expect)
2: Maybe we should not accept major re-factoring and leave it up to the discretion of thy maintainers to do so - aka Ian Jackson's option B. But there seems to be some disagreement around it. 

2.1: Some maintainers are concerned that they would have to deal with backporting, if we re-opened early.

2.2: A sensible compromise seems to me for the maintainer to evaluate whether a patch is ready to go in after RC3: if there is an amount of back porting that the maintainer can't deal with, it is IMHO OK for the maintainer to let the contributor know and give him/her the option to provide a patch for two trees as it is customary in Linux OR to wait until the ongoing release is out. This is a slight variant of Ian Jackson's option B in http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html <http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html> 

Of course, 1 will minimise the amount of incidents for 2

= VOTE =  
With that in mind: please vote on
A) Do we think 4.6 is in a good enough state to branch at the next RC (which would be RC3)

B) Do we have enough consensus given that there is some disagreement on how to deal with back-porting. In other words, does the proposal 2.2 above look sensible. 

Regards
Lars


[-- Attachment #1.2: Type: text/html, Size: 3083 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Vote] Re-open staging for contributions at RC3

[Andrew Cooper]

[-- Attachment #1.1: Type: text/plain, Size: 2768 bytes --]

On 02/09/15 17:12, Lars Kurth wrote:
> Hi all,
>
> I wanted to kick off a vote related to the following threads
> * http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html - [URGENT
> RFC] Branching and reopening -unstable
> * http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00543.html - [xen
> 4.6 retrospective] [urgent] rename "freeze" window and make release
> branch as soon as possible after RC1
>
> Although there was no consensus in the general case to say we should
> always branch at RC-x at any given release, there seems to be enough
> consensus for branching earlier, given a number of conditions are met:
>
> In particular:
> 1: We should not re-open staging too early (aka we would need to get a
> sense how much churn to expect)
> 2: Maybe we should not accept major re-factoring and leave it up to
> the discretion of thy maintainers to do so - aka Ian Jackson's option
> B. But there seems to be some disagreement around it. 
>
> 2.1: Some maintainers are concerned that they would have to deal with
> backporting, if we re-opened early.
>
> 2.2: A sensible compromise seems to me for the maintainer to evaluate
> whether a patch is ready to go in after RC3: if there is an amount of
> back porting that the maintainer can't deal with, it is IMHO OK for
> the maintainer to let the contributor know and give him/her the option
> to provide a patch for two trees as it is customary in Linux OR to
> wait until the ongoing release is out. This is a slight variant of Ian
> Jackson's option B
> in http://lists.xenproject.org/archives/html/xen-devel/2015-08/msg00883.html 
>
> Of course, 1 will minimise the amount of incidents for 2
>
> = VOTE =  
> With that in mind: please vote on
> A) Do we think 4.6 is in a good enough state to branch at the next RC
> (which would be RC3)
>
> B) Do we have enough consensus given that there is some disagreement
> on how to deal with back-porting. In other words, does the proposal
> 2.2 above look sensible.

My initial objections to branching early have been mitigated now, as it
turns out that the churn during the RCs is minimal, and the base quality
is substantially better than it has been in the past.

I don't forsee much else going in before the release, and there are no
huge blocker bugs needing working on (of which I am aware).

Therefore, I think it is now reasonable for work to start on -staging
again.  I don't forsee much more going into 4.6 before release. 
However, to avoid an invocation of sods law, some care should still be
taken.

Therefore, I think it is ok to branch and re-open rc3, but still use
maintainer prudence on more complicated series until a final rc has been
chosen, to reduce the risk of some awkward backports needing to happen.

~Andrew

[-- Attachment #1.2: Type: text/html, Size: 4763 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: [Vote] Re-open staging for contributions at RC3

[Ian Campbell]

On Wed, 2015-09-02 at 17:12 +0100, Lars Kurth wrote:
> Hi all,
> 
> I wanted to kick off a vote related to the following threads
> * http://lists.xenproject.org/archives/html/xen-devel/2015
> -08/msg00883.html - [URGENT RFC] Branching and reopening -unstable
> * http://lists.xenproject.org/archives/html/xen-devel/2015
> -08/msg00543.html - [xen 4.6 retrospective] [urgent] rename "freeze" 
> window and make release branch as soon as possible after RC1
> 
> Although there was no consensus in the general case to say we should 
> always branch at RC-x at any given release, there seems to be enough 
> consensus for branching earlier, given a number of conditions are met:
> 
> In particular:
> 1: We should not re-open staging too early (aka we would need to get a 
> sense how much churn to expect)
> 2: Maybe we should not accept major re-factoring and leave it up to the 
> discretion of thy maintainers to do so - aka Ian Jackson's option B. But 
> there seems to be some disagreement around it. 
> 
> 2.1: Some maintainers are concerned that they would have to deal with 
> backporting, if we re-opened early.
> 
> 2.2: A sensible compromise seems to me for the maintainer to evaluate 
> whether a patch is ready to go in after RC3: if there is an amount of 
> back porting that the maintainer can't deal with, it is IMHO OK for the 
> maintainer to let the contributor know and give him/her the option to 
> provide a patch for two trees as it is customary in Linux OR to wait 
> until the ongoing release is out. This is a slight variant of Ian 
> Jackson's option B in http://lists.xenproject.org/archives/html/xen
> -devel/2015-08/msg00883.html ;
> 
> Of course, 1 will minimise the amount of incidents for 2
> 
> = VOTE =  
> With that in mind: please vote on
> A) Do we think 4.6 is in a good enough state to branch at the next RC 
> (which would be RC3)

IMHO this decision should be the Release Managers to make. I'm not sure how
to vote to express that, so I suppose I'll abstain.

Since I'm sure the RM would want input from maintainers to help them make
this decision I'll note that my opinion (not vote) is that the tree is
indeed in a good enough state to branch.

> B) Do we have enough consensus given that there is some disagreement on 
> how to deal with back-porting. In other words, does the proposal 2.2 
> above look sensible. 

It's not clear to me which branch "ready to go in after RC3" refers to, the
reopened staging or the newly branched 4.6 branch?

If the proposal is to ask maintainers to use their discretion when applying
things to staging, i.e. taking the possible need to do backports (of
subsequent patches) to the 4.6 branch and/or asking for help with backports
from the submitters of patches which need it then that gets +1 from me.

Ian.

Re: [Vote] Re-open staging for contributions at RC3

[Lars Kurth]

> On 3 Sep 2015, at 10:15, Ian Campbell <Ian.Campbell@citrix.com> wrote:
> 
> On Wed, 2015-09-02 at 17:12 +0100, Lars Kurth wrote:
>> Hi all,
>> 
>> I wanted to kick off a vote related to the following threads
>> * http://lists.xenproject.org/archives/html/xen-devel/2015
>> -08/msg00883.html - [URGENT RFC] Branching and reopening -unstable
>> * http://lists.xenproject.org/archives/html/xen-devel/2015
>> -08/msg00543.html - [xen 4.6 retrospective] [urgent] rename "freeze" 
>> window and make release branch as soon as possible after RC1
>> 
>> Although there was no consensus in the general case to say we should 
>> always branch at RC-x at any given release, there seems to be enough 
>> consensus for branching earlier, given a number of conditions are met:
>> 
>> In particular:
>> 1: We should not re-open staging too early (aka we would need to get a 
>> sense how much churn to expect)
>> 2: Maybe we should not accept major re-factoring and leave it up to the 
>> discretion of thy maintainers to do so - aka Ian Jackson's option B. But 
>> there seems to be some disagreement around it. 
>> 
>> 2.1: Some maintainers are concerned that they would have to deal with 
>> backporting, if we re-opened early.
>> 
>> 2.2: A sensible compromise seems to me for the maintainer to evaluate 
>> whether a patch is ready to go in after RC3: if there is an amount of 
>> back porting that the maintainer can't deal with, it is IMHO OK for the 
>> maintainer to let the contributor know and give him/her the option to 
>> provide a patch for two trees as it is customary in Linux OR to wait 
>> until the ongoing release is out. This is a slight variant of Ian 
>> Jackson's option B in http://lists.xenproject.org/archives/html/xen
>> -devel/2015-08/msg00883.html ;
>> 
>> Of course, 1 will minimise the amount of incidents for 2
>> 
>> = VOTE =  
>> With that in mind: please vote on
>> A) Do we think 4.6 is in a good enough state to branch at the next RC 
>> (which would be RC3)
> 
> IMHO this decision should be the Release Managers to make. I'm not sure how
> to vote to express that, so I suppose I'll abstain.
> 
> Since I'm sure the RM would want input from maintainers to help them make
> this decision I'll note that my opinion (not vote) is that the tree is
> indeed in a good enough state to branch.

I am happy with Wei making that decision. But we chatted on IRC yesterday, and my interpretation of that conversation was that because of B) which is really a policy decision and because A) was raised specifically as an objection to Wei's earlier RFC, it would make sense to at least highlight this issue.  

>> B) Do we have enough consensus given that there is some disagreement on 
>> how to deal with back-porting. In other words, does the proposal 2.2 
>> above look sensible. 
> 
> It's not clear to me which branch "ready to go in after RC3" refers to, the
> reopened staging or the newly branched 4.6 branch?
> 
> If the proposal is to ask maintainers to use their discretion when applying
> things to staging, i.e. taking the possible need to do backports (of
> subsequent patches) to the 4.6 branch and/or asking for help with backports
> from the submitters of patches which need it then that gets +1 from me.

That's what I meant

> 
> Ian.

Re: [Vote] Re-open staging for contributions at RC3

[Stefano Stabellini]

On Thu, 3 Sep 2015, Ian Campbell wrote:
> > I wanted to kick off a vote related to the following threads
> > * http://lists.xenproject.org/archives/html/xen-devel/2015
> > -08/msg00883.html - [URGENT RFC] Branching and reopening -unstable
> > * http://lists.xenproject.org/archives/html/xen-devel/2015
> > -08/msg00543.html - [xen 4.6 retrospective] [urgent] rename "freeze" 
> > window and make release branch as soon as possible after RC1
> > 
> > Although there was no consensus in the general case to say we should 
> > always branch at RC-x at any given release, there seems to be enough 
> > consensus for branching earlier, given a number of conditions are met:
> > 
> > In particular:
> > 1: We should not re-open staging too early (aka we would need to get a 
> > sense how much churn to expect)
> > 2: Maybe we should not accept major re-factoring and leave it up to the 
> > discretion of thy maintainers to do so - aka Ian Jackson's option B. But 
> > there seems to be some disagreement around it. 
> > 
> > 2.1: Some maintainers are concerned that they would have to deal with 
> > backporting, if we re-opened early.
> > 
> > 2.2: A sensible compromise seems to me for the maintainer to evaluate 
> > whether a patch is ready to go in after RC3: if there is an amount of 
> > back porting that the maintainer can't deal with, it is IMHO OK for the 
> > maintainer to let the contributor know and give him/her the option to 
> > provide a patch for two trees as it is customary in Linux OR to wait 
> > until the ongoing release is out. This is a slight variant of Ian 
> > Jackson's option B in http://lists.xenproject.org/archives/html/xen
> > -devel/2015-08/msg00883.html ;
> > 
> > Of course, 1 will minimise the amount of incidents for 2
> > 
> > = VOTE =  
> > With that in mind: please vote on
> > A) Do we think 4.6 is in a good enough state to branch at the next RC 
> > (which would be RC3)
> 
> IMHO this decision should be the Release Managers to make. I'm not sure how
> to vote to express that, so I suppose I'll abstain.
> 
> Since I'm sure the RM would want input from maintainers to help them make
> this decision I'll note that my opinion (not vote) is that the tree is
> indeed in a good enough state to branch.

I also think it should be Wei's decision.


> > B) Do we have enough consensus given that there is some disagreement on 
> > how to deal with back-porting. In other words, does the proposal 2.2 
> > above look sensible. 
> 
> It's not clear to me which branch "ready to go in after RC3" refers to, the
> reopened staging or the newly branched 4.6 branch?
> 
> If the proposal is to ask maintainers to use their discretion when applying
> things to staging, i.e. taking the possible need to do backports (of
> subsequent patches) to the 4.6 branch and/or asking for help with backports
> from the submitters of patches which need it then that gets +1 from me.
 
+1

Re: [Vote] Re-open staging for contributions at RC3

[Wei Liu]

The tree is in good state as far as I can tell.

I asked Jan on IRC. He doesn't have strong opinion.

And from the feedback I got from this threads, several maintainers
expressed their opinion in favour of reopening the tree.

So I now make the decision we branch at RC3 and reopen xen-unstable for
new contributions.

I will send out an email to announce this.

Wei.