docs: sync ROADMAP/CURRENT_STATE with Native Runtime Phases A-C - #74
Conversation
Phases A (fail-closed admission boundary), B (live VRAM budget), and C (real telemetry) of docs/NATIVE_RUNTIME_V2_SPEC.md merged (#70, #72, #73) since these two files were last touched. Both had gone stale: - ROADMAP.md's Phase 3/4 rows described exactly the "remaining" work those three PRs closed (OrcScheduler wired into AdapterManager, telemetry surfaced) as still open. Corrected, and added a pointer to the new spec as the current foundation-hardening plan alongside the existing RUNTIME_PHASE0_SPEC.md contracts link. - CURRENT_STATE.yaml's native_runtime note predated all three phases. Added an accurate summary of what's landed, explicit that this is foundation hardening, not a default-runtime change (Phase D and the default-runtime flip remain open, per the spec's own scope). Docs-only, no code touched. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 33 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (3)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Four real accuracy gaps caught by an external review (Tools/grok-review.ps1 -Mode full): 1. CURRENT_STATE.yaml overclaimed Phase B as fully landed -- only the live VRAM budget READ shipped; the cost ESTIMATE side (OrcScheduler.EstimateRequiredBytes, still GGUF-file-size-only) was deliberately deferred in that same PR (#72). Corrected to name the gap explicitly. 2. ROADMAP.md's Phase 3 "Remaining" bullet had the same gap -- only named Phase D as open, omitted the deferred estimate work. 3. ROADMAP.md's "Last updated" banner (2026-07-17) was inconsistent with the 2026-07-19 status this PR stamps elsewhere in the same document. Updated the banner and clarified the doc's update policy allows incremental updates between releases, not just at release time. 4. NATIVE_RUNTIME_V2_SPEC.md's own banner still said "no implementation lands with this document" with no landed-phase status -- true in the narrow sense (implementation lands via separate PRs, exactly as designed) but misleading to a reader who'd reasonably read it as "nothing implemented yet." Added an explicit, dated implementation- status line naming which phases have landed (A/B-read-side/C) and which remain open (D, Phase B's deferred estimate half). Re-validated: YAML still parses, all 55 markdown anchor links (4 new) resolve correctly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
1 participant
What this is
Small, docs-only follow-up. Phases A (#70), B (#72), and C (#73) of
docs/NATIVE_RUNTIME_V2_SPEC.mdmerged sincedocs/ROADMAP.mdanddocs/CURRENT_STATE.yamlwere last touched, and both had gone stale.The staleness
ROADMAP.md's Native Runtime Phase 3/4 rows literally described the "remaining" work as: "keep proving the real-model path, surface SessionManager/AdapterManager-backed telemetry, wire OrcScheduler into AdapterManager" — this is the exact sentenceNATIVE_RUNTIME_V2_SPEC.md§0.1 quotes as its own founding mandate. Two of those three items (telemetry, OrcScheduler/AdapterManager wiring) are now done; only the real-model proof (Phase D) remains open.CURRENT_STATE.yaml'snative_runtimenote predated all three phases entirely.The fix
ROADMAP.md: corrected the stale "remaining" bullets in the Phase 3 and Phase 4 rows, and added a pointer toNATIVE_RUNTIME_V2_SPEC.mdalongside the existingRUNTIME_PHASE0_SPEC.mdcontracts link, distinguishing "contracts/design" from "current foundation-hardening plan."CURRENT_STATE.yaml: added an accurate, dated summary of what Phases A-C actually shipped, explicit that this is foundation hardening — not a default-runtime change. Phase D and the default-runtime flip remain open, exactly as the spec itself scopes them. Also bumped the file's ownlast_reviewed/reviewed_against_releaseheader fields.No overclaiming:
native_runtimestatus staysopt-in(unchanged) — only the note is corrected.Verification
git diff --stat: exactly 2 files, 24 insertions / 6 deletions.python -c "import yaml; yaml.safe_load(...)"after editing — still parses cleanly.docs/NATIVE_RUNTIME_V2_SPEC.md) confirmed to exist.🤖 Generated with Claude Code