chore(rivet): RQ-60-VFPPRESSURE implemented — 5/5 falcon stages verified, and why it stops short of verified - #1077
Merged
Merged
Conversation
…why NOT `verified` #1075 merged as 0ec9dc7 and did not flip its own status. SEVENTH instance of the class this session; #1076 (the status-evidence gate) is what makes this mechanical instead of dependent on me remembering. `implemented`, NOT `verified`, and the distinction is the point. The artifact's DoD says those symbols reach `nm -> T` **IN THE FUSED IMAGE**. What I verified, against the REAL ghcr components at 1.134.1 on cortex-m7dp — jess's flight core — with the failure established on main FIRST: iekf / position / attitude main exit=1 SKIPPED -> branch exit=0 rate / mixer already compiled, unchanged attitude#tick, position#tick reach `nm -> T` at 00000138 That is 5 of 5 PER-STAGE. The fused-image run has NOT happened — our local meld predates `fuse --pack-rebase` — and the rescued functions' correctness evidence is the fixture differential on the same code path (70/70 bit-identical vs wasmtime), not an executed attitude/ekf run. Earlier this session I flipped this same artifact to `implemented` PREMATURELY, reading "the PR merged" as "the stated outcome holds". The correction is not to flip more cautiously by feel — it is to use rivet's lifecycle for what it is: `implemented` = the code is in; `verified` = the evidence closes. jess's standing offer covers the fused image, INCLUDING if the answer is that it is worse on target. The reasoning is recorded NEXT TO THE DoD rather than only in this message, so the next person reading the artifact sees why the status stops where it does. v0.60: 3/8. Verified BY ID; CI-filter OURS=0; measured 473 = floor 473; claim_check exit 0. Refs #1069, Refs #1064 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YJK5LZZEkV5smCY1jKn18L
…MY flip #1076 landed the status-evidence gate, and on its first encounter with me it fired R2 on this very flip: FAIL R2 RQ-60-VFPPRESSURE: status `implemented` on a `manual:` done-when with no `verified-by` — 'the PR merged' is not 'the stated outcome holds'; write the basis down It was right. My flip WAS justified — I verified 5/5 against the real ghcr components — but I wrote that reasoning as PROSE in a status note beside the DoD, where no machine can read it. A human-readable caveat and a machine-checkable claim are not the same artifact, and the gate exists to collapse that gap. The basis is now in `verified-by`, stated so it can be DISPUTED rather than trusted, and deliberately not overstated — the gate forces the basis to be written and cannot judge it, so inflating it here would defeat the mechanism I just commissioned: * per-stage, NOT fused: real ghcr components at 1.134.1, failure established on main FIRST, then the same bytes on the branch at -t cortex-m7dp — iekf/position/attitude exit=1 SKIPPED -> exit=0, rate/mixer unchanged, attitude#tick and position#tick at `nm -> T` 00000138 * mechanism soundness is the fixture differential on the same code path (70/70 bit-identical vs wasmtime), NOT an executed attitude/ekf run * THE FUSED-IMAGE RUN THIS done-when NAMES HAS NOT HAPPENED — local meld predates `fuse --pack-rebase`; jess holds that evidence under their standing offer, and `verified` waits on it INCLUDING if the answer is worse on target status-evidence exit 0, claim_check exit 0, CI-filter OURS=0. Refs #1069, Refs #1064 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YJK5LZZEkV5smCY1jKn18L
avrabe
force-pushed
the
chore-vfppressure-flip
branch
from
August 27, 2026 05:21
95e0f1b to
8f543f1
Compare
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
#1075 merged and did not flip its own status — the seventh instance this session. #1076 is what makes this mechanical.
Verified against the real components, not the lane's report
Pulled
iekf/position/attitude/rate/mixerat 1.134.1 from ghcr and established the failure onmainfirst:attitude@0.7.0#tickandposition@0.7.0#tickreachnm → Tat00000138. On cortex-m7dp — jess's flight core, not the easier single-precision target.Why
implementedand notverifiedThe DoD says those symbols reach
nm → Tin the fused image. I verified per-stage. The fused run hasn't happened — our local meld predatesfuse --pack-rebase— and the rescued functions' correctness evidence is the fixture differential on the same code path (70/70 bit-identical vs wasmtime), not an executedattitude/ekfrun.Earlier this session I flipped this same artifact to
implementedprematurely, reading "the PR merged" as "the stated outcome holds". The correction isn't to flip more cautiously by feel — it's to use rivet's lifecycle for what it is:implemented= the code is inverified= the evidence closesjess's standing offer covers the fused image, including if the answer is that it's worse on target. The reasoning is recorded next to the DoD in the artifact, not just in the commit, so the next reader sees why the status stops where it does.
v0.60: 3/8. Verified by id,
OURS=0,measured 473 = floor 473, claim_check exit 0.