Every GitHub issue closed before the tracker migration on 2026-08-14 — Waves 1 through 3 and the
release-repair issues between them. 17 closed issues, verified with gh issue list --state closed --limit 1000 (the default limit is 30; the count here is the one actually returned).
These are not re-imported as backlog tasks, deliberately. Backlog IDs follow creation order, so a
VQR-00NN could never be made to match the #NN already cited in commit messages, wave reports and
issue cross-references — importing them would create a second ID space over the same history, and 17
Done rows would compete with the board’s only real signal, which is what is left.
The issues themselves were deleted from GitHub on 2026-08-14. gh issue view <N> no longer
resolves and the #NN URLs 404. The bodies and all 26 comments live in archive/issues.json,
committed and pushed before the deletion, with lab identifiers replaced by stable placeholders —
archive/README.md carries the mapping and the verification:
jq -r '.[] | select(.number==11) | .title, "", .body' archive/issues.jsonThe full narrative — including every non-green CI run and its disposition — is in the gitignored
codex/report-20260812-wave1.md and codex/report-20260812-wave2.md.
Numbers absent from this table (#1, #14, #19, #23) are pull requests, not issues and not archived here. #16 was Renovate’s Dependency Dashboard, bot-owned and never a task; it was deleted with the rest and Renovate will recreate it at a new number on its next run. #20 was the only open issue and is now VQR-0001 on the board — it was deleted too, so the task is the live record and the archive is the original.
Resulting SHAs are given where they are derivable. Commit messages in this repo do not cite issue
numbers, so the mapping below is by content — file-addition history and the wave reports — not by an
explicit Closes #N. Where a lane’s work was squashed into the Phase 1 integration commit, that is
said rather than guessed at.
| # | Title | Closed | Resulting SHA(s) |
|---|---|---|---|
| 2 | Build and deploy first working RFC 6035 to OTLP release | 2026-08-12 | 25f9b4d, 8763e9c, ce49568, 206c6f1 (all of Wave 1) |
| 3 | Wave 2: sibling parity, public signal contract, and v0.1.0 | 2026-08-13 | umbrella; released at bd0168b (v0.1.0) |
| 4 | P1-L1: Correct the public metric and log contract | 2026-08-12 | d05f2d7 |
| 5 | P1-L2: Add bounded collector self-observability | 2026-08-12 | 4d264c6 |
| 6 | P1-L3: Add release-please and GoReleaser plumbing | 2026-08-12 | 3bcb027, 973853c (initial-version: 0.1.0) |
| 7 | P1-L4: Port and harden the complete workflow set | 2026-08-12 | efb5247 |
| 8 | P1-L5: Write operator documentation and community files | 2026-08-12 | fda5313 (docs, CONTRIBUTING.md, SECURITY.md first appear there) |
| 9 | P1-L6: Generate the signal catalog, dashboard, and alerts | 2026-08-12 | fda5313 (spec/signal-catalog.json, dashboards/, alerts/ first appear there) |
| 10 | P1-L7: Add the community Helm chart | 2026-08-12 | fceae20 |
| 11 | P2-L8: Prove both dialects and deduplication in a live soak | 2026-08-13 | 4df6e82 |
| 12 | P2-L9: Prove end-to-end docs hub publishing | 2026-08-12 | a9521cb |
| 13 | P2-L10: Fuzz the parser from real and RFC seeds | 2026-08-12 | c252715 |
| 15 | GATE integration: reconcile Phase 1 and first-run CI | 2026-08-12 | fda5313 |
| 17 | test: synchronize SIP observer assertions | 2026-08-12 | 8921f62 |
| 18 | fix: preserve the frozen build-info Prometheus series name | 2026-08-12 | 1259341, released f64160d (v0.1.1) |
| 21 | Wave 3: operational Dashboard v2 and Grafana reconciliation | 2026-08-13 | 951cc2d, ccd0d60, fae5290, 66935b3, 23731cb; released c0e08eb (v0.2.0) |
| 22 | Design: assess binary RFC 3611 RTCP-XR receiver support | 2026-08-13 | design assessment, no implementing commit |
The two facts from this history a future session will otherwise re-derive
The build-info defect (#18) is why the signal contract is treated as frozen. An explicit OTel
gauge unit of 1 made the SDK emit rfc6035_2otel_build_info_ratio. v0.1.0 was not retagged; the fix
shipped additively as v0.1.1.
1.0.0 is blocked on two observations, neither of which has happened: a non-Poly RFC 6035 sender ingested, and the pre-standard grammar seen from a second Poly model or firmware. Recorded at the close of Wave 2. Do not plan a 1.0.0 wave until both are observable.