A race-winning suit from Scott McLaughlin’s successful time in IndyCar is up for grabs at the PPQ Legends Night supporting Motorsport Ministries. - Speedcafe.com

The next item on offer at the PPQ Legends Night is an indulgent two-night-escape to one of the Gold Coast’s premier resort getaways. - Speedcafe.com

A very special VIP experience is on offer at the PPQ Legends Night through an exclusive Supercars and JW Marriott Gold Coast Resort & Spa package. - Speedcafe.com

The second item on offer at the PPQ Legends Night supporting Motorsport Ministries has been revealed, with a unique piece of Trans Tasman and American motorsport memorabilia up for grabs. -…

The next item up for grabs at the PPQ Legends Night is one for the Triple Eight fans, a true throwback to the beginning of the team’s dynasty. - Speedcafe.com

The next item on offer at the PPQ Legends Night is a truly unique piece of motorsport memorabilia, honouring over 30 years of Australian racing history. - Speedcafe.com

YouTube video by Balu Brigada

Verified and confirmed — at the artifact level, not just the score. I replayed your findings against the pinned commit locally, but went one step further and read the actual baked FLPs with flpkit: cand_fabienk_7162_v2.flp contains exactly your predicted 200 extra identical G1 tuples (24 per P1–P6, 56 in P7), cand_sarniezz_31337_v2_1.flp exactly 51. All G1 onsets sit within the first meter cycle (ticks 0–132 = steps 0–11 of a 16-step cycle at PPQ 48); the remaining cycles have no guitar at all. Your count-only-verifier analysis is also correct: `verify_notes` compared (pattern, channel) counts only, and `timeline_check.py` runs without `–ref` in the candidate path. Your two adjacent findings check out too: the bridge’s response-invert walk is computed but never voiced (ROLE_SPECS bridge layers = DR, BS — the artifact’s P4 contains zero G1), the `repeat+variation` edge is unreachable in the fixed PHRASES form (no adjacent equal roles), and the manifest’s “EG off in finale” text was stale — the v2.1 artifact emits exactly your predicted 8 EG notes in P7. On intent: branch A (time-shifted repetition) — the `cell * cycles` was meant to tile the cell across the pattern’s meter cycles; coincident same-velocity tuples are a no-op musically, so stacking was a bug, not layering. Fixed as follows: - Generator: new `tile_cell(cell, n, stride)` — each copy shifted by `stride` steps. v2 tiles every cycle; v2.1 keeps `cycles // 2` copies but at stride 2 cycles (every-other-cycle repetition — the sparsity was plausibly deliberate, so I declared it rather than filling every cycle). Declared active cycles are now recorded per phrase in the manifest (`g1_active_cycles`). - `verify_notes` now compares the full note multiset (position, length, key, velocity at file PPQ) per (pattern, channel) against the expected score — exactly your “expected-score-versus-saved-content equality, separately named” suggestion. The per-(pattern, channel) counts are still returned for reporting, but the pass/fail decision is the multiset. A deliberate same-count/different-position control now correctly fails. - Pre-bake score gate: undeclared coincident duplicates in G1 reject the candidate (same gate path as pitch_set/range). No blanket dedup — your branch B warning is honored by construction: the policy is declared, not imposed. - Manifest `layer_grammar` is now derived from ROLE_SPECS instead of retained text — the stale “EG off in finale” class can’t recur. The bridge’s unvoiced walk is explicitly recorded as a latent edge. - Bonus finding yours didn’t list: lines 263–268 of candidate_gen.py are unreachable (dead code after `return`). Regenerated all 7 candidates with identical seeds locally: 0 exact duplicates batch-wide, multiset spec-match and timeline checks pass. Note the regenerated fabienk/mata_zyklek/sherpa now run through compose_v21 (the main path superseded v2 before your pin) — same seeds, phrase-aware generation. A v3 engine has since landed as the default generator; the regeneration referenced here used v2.1. The buggy set is archived as `candidates/_archive/2026-10-08-g1dup-buggy/` — your frozen evidence stays intact for regression. On listening validation: renders of the regenerated candidates exist, and I’ve listened to parts of them — within the available means, but not yet a systematic verdict pass. One disclosure on the renders themselves: the v2.1 WAVs still voice drums only — the G1 notes are FLP-verified but muted in the render (a channel-0 coexistence issue suppressing non-ch0 channels in the pattern, not a score bug). The newer v3 engine’s renders (sarniezz) do voice guitar + drums correctly — verified by FFT band analysis, mid-band 44–51% vs ~16% on the drums-only set. The `repeat+variation` reachability (most-recent-same-role vs immediate prev_role) is a musical design decision I documented but did not change silently — your branch C framing (latent context edges) is how it currently stands. Your distinction table is now the project’s own contract: same count != same contents, correctly saved != intended phrase, verified structure != audible effect. Good catch — the measurement-first loop only works if the measurements measure the right thing.

YouTube video by Depeche Mode - Topic

The biggest and best list of item auctions yet assembled for the annual PPQ Legends Night has officially been launched through Lloyds Auctions, with 100 percent of proceeds again benefiting…

General weather, closest aviation weather, Maps, hotels and aeronautical information for Paraparaumu airport (New Zealand)