docs: correct migration plan after #516/#517 - #518
Merged
Conversation
…parked not superseded - Phase 2 row 3 no longer owns the sunrise/sunset crossover. #517 showed that period was never a tie: the action set was empty (every lattice candidate overshot the 86 W residual, #497 excluded the unexecutable ones), so IDLE won by default. P3 subsumed the case P2 was to handle. Row 3's acceptance is now the genuine near-ties (#466 evening, #393). - Phase 4 gains the crossover regression cover and #352's residual band. - #354 reclassified: NOT 'already fixed'. #511 removed sub-material exports from the plan; #354's own 0.1-0.5 kWh home-dominant band survives. Its mechanism is the wrong layer (post-hoc demotion = P>R optimism), but its two-sided materiality test is carried into Phase 4 as domain knowledge. - Phase 0 items ticked with what actually happened, including that the fold-scoping premise check FAILED and redirected the design to #511. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Plan-doc corrections only — no code. Three substantive changes, plus ticking Phase 0 against what actually landed.
1. The sunrise/sunset crossover moves from Phase 2 to Phase 4 (#517).
The plan specified ridax67's 06:00–06:59 period as the acceptance test for dropping Phase 2's
epsilon <= 0early-return. #517 established that period was never a tie at all — the action set was empty (every lattice candidate overshot the 86 W residual, and #497 correctly excluded the unexecutable ones), so IDLE won by default rather than by a coin flip. P3 (candidates are executable commands) subsumed the case P2 was going to handle. Left uncorrected, a Phase 2 implementer would test something already fixed elsewhere and could wrongly conclude the tie policy caused it. Row 3's acceptance is now the genuine near-ties: #466's evening periods and #393.2. #354 is parked, not superseded — and explicitly not "already fixed".
#511 removed sub-material exports from the plan, but #354's own body flags the 0.1–0.5 kWh home-dominant band that survives it: material exports the DP legitimately plans that still commit the inverter to
grid_firstand forfeit load-following headroom. #352 stays open for that residual. #354's mechanism is the wrong layer (post-hoc demotion means the DP already credited revenue the mapping forgoes — the P>R optimism P3/P4 forbid), but its two-sided materiality test is real domain knowledge now recorded for Phase 4: a near-full-rate export has almost no headroom left to protect, so a home-dominance-only rule would permanently demote every export on a high-consumption house.3. Phase 0 ticked against reality, including that
test_flow_coherence.pywas already on main (PR #506, 2026-08-08 — the plan's "not on main" note was wrong), and that the fold-scoping premise check failed and redirected that work into #511.Note
This is the second time the plan doc has caught a forward-regression before it shipped (the first was #516's row-2 parity claim). Both were found by a PR contradicting a normative claim rather than by a user report later.
🤖 Generated with Claude Code