stale PRs


(Drew Crawford) #1

Hello all,

I'm growing concerned about a small workflow hangup.

I've noticed a pattern in the PR queue where: a patch is generally uncontroversial, reviews are of very high quality, and authors are quick to respond to reviewer concerns. However so much time elapses waiting for review or re-review that the patch becomes stale. Authors bump infrequently to try not to be annoying, and I am sure reviewers grow weary as the bumps are a lot louder when there are 80 people doing it.

Examples I've been following:

https://github.com/apple/swift/pull/629
https://github.com/apple/swift/pull/1176
https://github.com/apple/swift/pull/1479
https://github.com/apple/swift/pull/1891
https://github.com/apple/swift-corelibs-libdispatch/pull/61
https://github.com/apple/swift-corelibs-libdispatch/pull/62
https://github.com/apple/swift-corelibs-libdispatch/pull/64

Speaking only for myself, dreading this outcome has prevented me from sending ~10 patches upstream in the last two months. I can work around a bug from my code in five minutes; but to prosecute that fix upstream and into a snapshot takes 6 weeks, and it is just not worth the effort to me the vast majority of the time. Only in cases where I absolutely must, and even then, the experience resolves me to do it locally next time. Meanwhile, I wonder if some of the very bright reviewers I've had the pleasure to work find pouring over a mountain of boring patches involving dark corners of platforms they don't personally use and problems they don't have to be a source of joy.

Now, perhaps this is all by design, and if so: Okay. But if it isn't by design, I wonder if there is some small change we could make to increase everyone's happiness.

Respectfully,

Drew