Completing the Review Cycle
Introduction
Welcome back to Comments That Get Merged. You’re at Lesson 5, the complete review-cycle milestone, where we’ll bring the review skills we’ve practiced together in one complete pull request cycle. We’ll follow a ticket archival change in the order it happened: the requirement and original proposal, the initial review, the author’s fix, and a careful re-review. By the end, we’ll know how to support a request for changes with clear evidence, then approve only after checking that the blockers are resolved.
Reading the Requirement and the Original Proposal
Before judging a change, we compare it with the requirement. In INITIAL_DIFF.md, the rule is clear: a ticket may be archived after it is closed, but an open ticket must not be archived. The original proposal adds an archive method without checking the ticket’s status:
Because this method accepts every ticket, an active ticket can be marked archived before its work is finished. In a real support system, that could make unfinished work disappear from normal views. This is a lifecycle violation, not a style preference, so it needs a blocking finding. As in earlier lessons, we start with the agreed-upon requirement and describe the concrete impact.
The proposal also includes a test that archives a newly created ticket, which is open by default, and expects the operation to succeed. That test does not just miss the boundary: it accepts the behavior the requirement forbids.
Assembling a Coherent Request Changes Review
As in earlier lessons, we read the review as a timeline: Original proposal → Reviewer request → Author revision → Verified resolution. The initial review below is written against the original proposal shown above, so its present-tense findings describe that proposal, not the revised code we’ll look at next.
A complete review in REVIEW.md explains the available review states: Comment for feedback without unresolved blockers, Request changes when a blocker remains, and Approve when the current revision has no unresolved blockers. The initial review identifies two distinct blockers:
The review also includes an optional nit about status constants and praise for keeping archival state beside ticket lifecycle state. Those comments add useful context, but neither cancels the two blockers. The requested changes are specific: add a closed-status guard, and test both successful archival after closing and rejection while open. Since merge risk remains, Request changes is the accurate choice; comment-only would understate the problem.
