On Fri, 16 Jan 2015 11:23:02 -0800 Walter Bright via Digitalmars-d <[email protected]> wrote:
> On 1/16/2015 9:49 AM, Andrei Alexandrescu wrote: > > I'm thinking of something like: if there's $(legitimate) request for > > changes but > > the author is dormant for more than $(X) days, then close. > > That's also a hamfisted policy. I've seen PR's that were good, but needed a > bit > of work, but the author disappeared. Sometimes, I've taken those over and > finished them. > > Arbitrarily closing them means they get lost forever. that's why there will ALWAYS be alot of items in queue and it will not be processed at any sane rate. add ten more people to process the PR queue, and... there will be just more and more unprocessed PRs. it's cause you're looking at that queue like it's a backpack full of things that can lay there forever. sometimes somebody can pull some thing out of backpack and work on it. but backpack will accumulate more and more things over the time. this will continue unless you realise that PR queue is not a backpack full of possible cool things. it's a queue of things that *must* *be* *processed* *within* *short* *time*. if it took more than ten days to simply react on the thing -- throw it out, it just cluttering the queue. this is one of the things that "github culture" gets very-very wrong. a proper place to accumulate requests and code is bugzilla. and PR queue is a queue for things that finally moved to "merging stage". i.e. for things that must be reviewed and merged (or rejected) ASAP. it doesn't matter how hard somebody will try to keep that queue "manageable", it will be oveflown with requests. the only way to manage it is to turn it from backpack to "ASAP-queue". so i repeat my point: if something sits there for twenty days (let's add ten days) without any motion -- remove it from the queue. either this, or it always be cluttered.
signature.asc
Description: PGP signature
