Pick an option on each card and press Done; your answer goes straight
back to the session. The raiser's pick is marked. "I'm not sure" is a real answer.
1
In plain terms. Should we drop the last small polish to the move-in cost checker, or review it once more and merge? It catches a few more made-up wrong answers but now fails 4 unusual correct wordings; none of Clara's real answers are affected either way, and your original fix is already merged.
The full question, as raised
LANE AUDIT deposit-eval-grader row 3: PR #17829 at head 31341354 closes several crafted double-count leaks but newly fails 4 crafted correct wordings vs main (disclosed); 0 of 96 real replies exercise the changed rules. Gera's bar is no flaky tests. Close it and cancel the row, review once more then merge, or merge as-is?
Why, and what was tried
The session raising this recommends: Drop it; the main fixes stay merged Fable split on order between dropping and one more review; both agree merging as-is is out, and both note the PR misses its own bar: it newly fails 4 correct wordings (e.g. 'Your application fee goes toward it.') and Fable found no shape that separates them from the double counts it catches, so another review can't fix that. You asked for no flaky tests. 0 of 96 real replies touch these rules, and your original ask (#17646) plus the redesign (#17788) are merged. Reversible: a closed PR can be reopened. Model receiptFABLE was asked, and escalated. Receipt f7a174109 — NEEDS HUMAN. THE TWO PASSES DISAGREED, AND ONLY THE OPTION ORDER DIFFERED. as-given → '(b) One Fable review at 31341354, then merge if nothing new goes the wrong way' (medium) reversed → '(a) Close #17829 unmerged and cancel row 3; record the cousin wordings as known limits' (medium) An answer a reordering can flip is not an independent judgement, it is position bias wearing one. Escalating with both readings attached.
as-given reasoning: ROW.md states the row's own merge bar: '…every main->head flip goes the right way, each new rule pinned by a row, Fable review at head'; STATUS.txt shows hold-for-…
4eae1e31 is parked on this.
Answered and withdrawn (3)
b91422849 Should we merge the fix that stops the move-in cost test from failing Clara when she correctly says two applicants pay $40 each? Merging ends those false failures but lets one unusual wrong answer slip through; tightening more brings some false failures back. — withdrawn
b91424382 Can the helper add a number of applicants to each test case, so the checker for Clara's move-in cost answers stops guessing it from her wording? Yes ends the leaks two reviews found; no means we keep patching wording and the fix stays unmerged. — answered: Yes, add the applicant count to the test data
b91442961 Should we stop polishing the move-in cost checker and drop the last small fix? It catches a few made-up wrong answers but would start failing several correct ones, and none of Clara's real answers are affected either way; the main fixes are already merged. — withdrawn