[dmd-beta] Cherry-picking II

[Andrew Edwards]

On 2/2/14, 2:04 PM, Jesse Phillips wrote:
> Well, my opinion is that the quest to find a "simple" process is 
> leading to reinventing the wheel with a fork and a spoon rather than a 
> hammer and chisel. This new issue is not a new problem and would have 
> been addressed with the appropriate process of creating pull requests 
> against the release branch, or even merging to the release branch 
> instead of master. But there seems to be some confusion about this 
> (now that it has come up as a replacement for writing notes back and 
> forth).
>
I agree with you that what we are doing now complicates matters more. It 
is not my choice that I made the decision to cherry-pick commits to the 
release branch, rather it is compromise that promotes forward progress. 
Fact is, only three people adhered to the request to create pull 
requests against the release branch and there are a lot more 
contributors than that. Honestly I believe this is the best approach but 
no matter how good one things a particular approach is, if no one is 
doing it, it serves no purpose.

> In your last email you took issue of adding a second review step, when 
> the first doesn't get any attention. This is wrong, dead wrong. There 
> should not be two reviews, there should only ever be one single pull 
> request and that request should get the review.
>
I took issue with it because by adding this extra step, we are inducing 
additional delay in the entire process. As you said, one review should 
suffice. All that's left to do after that is flip a switch indicating 
that it needs to be in the release branch. This can be accomplish in one 
of two ways: 1) set the Asignee to AndrewEdwards following the merge; 2) 
Set the Milestone to 2.065 anytime during the creation, review or merger 
of the request. Come to think of it, a combination of the two would 
probably provide the best solution. The author assigns the milestone 
when creating the pull. Upon seeing this milestone the reviewer assigns 
the pull to me after the merge. A two step process that requires one 
additional step from two different people.

> Yes, this approach does require involvement of the patch contributors, 
> but that is already true. The contributor is already tackling 
> regressions for the new release, so they're already aware of where 
> this fix needs to go. It just requires them to check out the release 
> branch on the three repos before starting.
>
Though I'd like this to happen, I really doubt it is going to. 
Therefore, in order not to prolong the process, I've decided not to 
pursue this direction any longer.

> Even if the developers don't wish to push the request against release, 
> they can still leave a comment on the Pull request (against master) 
> stating this should go into release. At this point the pull request is 
> NOT merged into master, instead it is merged into the release branch. 
> This approach has the negative that the fix may not be complete or 
> have merge conflicts due to dependency on previous changes (this would 
> be where a comment should be left and request that the issues be 
> addressed).
>
> Once the release is complete, merge it back into master (this can 
> actually be done at any time, but should always be done at the end of 
> release).
>
>
> On Sun, Feb 2, 2014 at 9:29 AM, Andrew Edwards <edwards.ac at gmail.com 
> <mailto:edwards.ac at gmail.com>> wrote:
>
>     So this is where the discussion ends? No decision or further
>     communication. All forward action suggest the status quo. We
>     honestly need to do something here gents. Were it my decision to
>     make, would have been handled on day one but since it's not, I
>     need your participation.
>
>     Kenji, need you take a look at dmd/pull/#3140 and see what kind of
>     problems will occur it is merged after #3103, #3151, #3174, and
>     #3169 respectively.
>
>     Thanks,
>     Andrew
>
>     _______________________________________________
>     dmd-beta mailing list
>     dmd-beta at puremagic.com <mailto:dmd-beta at puremagic.com>
>     http://lists.puremagic.com/mailman/listinfo/dmd-beta
>
>
>
>
> -- 
> Jesse Phillips
>
>
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-beta/attachments/20140202/793d6aba/attachment.html>