[Dlang-internal] The Phantom Zone

[Brad Roberts]

On 1/15/2018 7:50 PM, Martin Nowak via Dlang-internal wrote:
> On Monday, 15 January 2018 at 01:10:58 UTC, Walter Bright wrote:
>> But github has a feature we can use for this. We tag these PRs with 
>> "Phantom Zone" and then close them.
>>
>> http://superman.wikia.com/wiki/The_Phantom_Zone
>> (People with terminal diseases would get put in the Phantom Zone 
>> until a cure could be found.)
>>
>> They can then be viewed via this link:
>>
>> https://github.com/dlang/dmd/pulls?q=is%3Apr+is%3Aclosed+label%3A%15Phantom+Zone%15 
>>
>>
>> PRs banished to the Phantom Zone would have to have an explanatory 
>> comment saying why.
>>
>> Andrei has suggested calling it "Limbo".
>
> What other projects are occasionally doing is a mass exodus of all 
> (stale) PRs. Asking people (via comment) to create new ones (hurdle) 
> if they still care enough.
> I've seen this many times, sometimes because repos are migrated (good 
> reason), sometimes just because it's too many.
>
> I'd suggest that someone plans a friendly spring-cleaning with a 
> helpful message. It won't be very disturbing IMO.
> Making sure to get all the right stalled PRs is another task. Maybe 
> just take everything > 12 month or so. Some research into PR stats 
> would be helpful, e.g. likelihood of merge after X days, to inform a 
> good value.
>
> Tagging closed PRs as phantom zone is a good idea.
> Let's just no do it continuously, but as properly communicated mass 
> cleaning.
>
> Overall this sounds like a couple of bash lines, candidate for 
> https://github.com/MartinNowak/github_scripts.

What I'd like to see as a first step is to draw a line in the sand, 
every new pull request as of 1/1/2018 should be handled to resolution, 
period.  Don't let them get stale.  If really necessary, maybe taking 
the PZ action, but that'd really be more of a failure than success, but 
less so than just letting it age into the lower depths of the queue.

Hopefully that first step can be taken and successfully pulled off. If 
so, then some effort can be put towards shrinking the backlog. But I 
haven't seen any real evidence out side of short flurry of activity that 
we can even keep up.