[Trying -hackers rather than -www this time, since the impacted users are here.]
There are occasionally patches that are dutifully rebased by a
responsive author, all feedback implemented... but there have been no
reviews for a while, and there's no sign of any on the way. This case
seems to tie us up in knots. We don't want to Reject since the patch
is fine, it's just not a current priority; and we don't want to Return
with Feedback because there is no feedback to act upon. Since no one
wants to close it, they can drag on forever, with hopeful authors
rebasing eternally. I ended up closing several patches like this with
RwF, but I felt the need to write a huge explanation in the
accompanying email.
This has been discussed before (e.g. [1]); the two competing proposals
I've seen are to add a new close state or to simply remove the "with
Feedback" and treat all Returned patches the same. I can see the case
for minimizing the number of choices, but in this patchset, I've opted
to implement a new state. I think it's useful to communicate to a
contributor that their next steps, if they still want to pursue them,
need to be focused on coalition building rather than code changes. And
I think Returned with Feedback has a useful meaning already.
0001 just adds the "Returned: Needs more interest" state to the
database and makes it available to the UI. The (optional) 0002 goes
farther, and puts the two Returned states along with the Next CF state
into a new "Deferred" section in the UI. (It's UI-only; the app still
treats them as closed patches for the purposes of the CF.) This
emphasizes the distinction between Returned and Rejected: Rejected is
meant to be a full stop to the story, while Returned is more of a
comma.
Screenshots attached. WDYT?
--Jacob
[1] https://www.postgresql.org/message-id/flat/3905363.1633288498%40sss.pgh.pa.us
