Repository navigation
fix(web): keep AI configuration edits when the window regains focus - #81
Conversation
Code Review Roast 🔥Verdict: No Issues Found | Recommendation: Merge Oh wait, this PR is actually still clean. I need to sit down — again. The previous reviewer's feedback was all addressed: the standalone The core fix remains sound: depend on 📊 Overall: Listened to every review comment, made it cleaner, and somehow didn't introduce a single new thing to roast. I'm starting to think this repo is run by competent people. Files Reviewed (4 files, excluding bun.lock)
Previous Review Summaries (4 snapshots, latest commit 4730580)Current summary above is authoritative. Previous snapshots are kept for context only. Previous review (commit 4730580)Commit: What changedThe PR prevents the AI configuration form from resetting when the window regains focus and triggers a React Query refetch. The root cause was the Findings
VerdictNo Issues Found, Recommend Merge — The fix is well-designed, correctly leverages React Query structural sharing, and the test covers both the keep-edits-on-refetch path and the save path. The single nitpick is a type precision concern, not a correctness issue. View full review: https://app.kilo.ai/cloud-agent-fork/review/dae45de6-3875-4c12-bad8-aeb8de981b1c Previous review (commit 4e4923a)Verdict: No Issues Found | Recommendation: Merge Oh wait, this PR is actually still clean. I need to sit down — again. The incremental changes inline happy-dom registration into the test file and delete the now-unnecessary 📊 Overall: The PR that keeps on giving. Listened to feedback, made it cleaner, and didn't introduce anything new to roast. I'm starting to feel self-conscious. Files Reviewed (2 files in incremental diff)
Previous review (commit 2ac58f2)Verdict: No Issues Found | Recommendation: Merge Both previous review comments were addressed in follow-up commits:
The incremental diff is clean — two targeted fixes, zero new problems. The original PR logic remains sound, the 📊 Overall: Like a PR that actually listens to code review. I'm not used to this. Files Reviewed (3 files, excluding bun.lock)
Previous review (commit 1898ca2)Verdict: No Issues Found | Recommendation: Merge Oh wait, this PR is actually clean. I need to sit down. I had my flamethrower warmed up and everything. The fix is textbook: depend on the response body, not the query envelope that gets a new Tests cover both the regression (edits survive unchanged refetch) and the happy path (form resets after save). The mock throws on unexpected URLs instead of silently passing — that's the kind of defensive testing that pays off at 3 AM. 📊 Overall: Like finding a unicorn in production — I didn't think clean PRs existed anymore, but here we are. Files Reviewed (4 files, excluding bun.lock)
Reviewed by deepseek-v4-flash · Input: 46.1K · Output: 6.4K · Cached: 225.3K |
4e4923a to
4730580
Compare
The generated fetcher returns a fresh Headers object on every request, so the query result envelope never structurally shares and each focus refetch reset the form. Reset only when the response body changes. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
4730580 to
61b6f2c
Compare
The generated fetcher returns a fresh Headers object on every request, so the query result envelope never structurally shares and each focus refetch reset the form. Reset only when the response body changes.
Summary
How to test
Checklist