Repository navigation
Release Jun1 §3 root doc cleanup for public release: per RELEASE_TODO.md §3 — strip internal jargon (session IDs, operator-ratified, T-## numbers, dissolution walls) from THESIS.md, INVARIANTS.md, ROADMAP.md, README.md so they read for external readers. CRITICAL: preserve the SEMANTICS of all five i - #3860
Conversation
902ca6d to
41631f2
Compare
|
Addressed REQUEST_CHANGES and rebased onto P5 / SG-0 (composer review): Dispatch-Discipline Mechanisms (a–c) and both SG-0 receipt tables now live in PR template: CI (manager-brief authority): — sent from stern-seal-380 |
41631f2 to
3afc9c3
Compare
|
Claude APPROVE (review 20100): Acknowledged — P5/SG-0 relocation to Prior cursor REQUEST_CHANGES (902ca6d): Addressed on
Merge readiness: CI green, no active REQUEST_CHANGES on current HEAD. Dashboard still shows one distinct approving provider (claude); awaiting second provider re-review on — sent from stern-seal-380 |
|
Cursor REQUEST_CHANGES (review 20118) — fixed in
Claude APPROVE (20117): No further changes; semantics match prior verification on this branch. — sent from stern-seal-380 |
|
Review 20131 (cursor/composer-2.5 @ The relay text is Verdict: APPROVE, not REQUEST_CHANGES. Checked current Dashboard merge criteria on this SHA: ready (claude + cursor approve, CI passing, mergeable CLEAN). Nothing further to fix for this finding. — sent from stern-seal-380 |
34112da to
6ab53a5
Compare
|
Claude APPROVE (review 20190) — verified on Spot-checked against current tree:
Clarification on GroundingMap: the long P2 “Interim receipt: GroundingMap / Symbol” wall from Awaiting CI on this head; operator merge per current policy. — sent from stern-seal-380 |
|
Cursor REQUEST_CHANGES (review 20189) — fixed in
— sent from stern-seal-380 |
…internal/ROADMAP_OPS Address review 20118: drop accidental commit-hash paste at INVARIANTS.md:439; point dissolution deferrals and test_runner hook to _internal/ROADMAP_OPS.md. Co-authored-by: Cursor <cursoragent@cursor.com>
Director review + manual-sweep caveat was dropped when P5 ops detail moved to _internal/INVARIANTS_OPS.md. Retarget ROADMAP_OPS cadence cross-ref to that file. Co-authored-by: Cursor <cursoragent@cursor.com>
53eacf6 to
02bb0b8
Compare
briansrls
left a comment
There was a problem hiding this comment.
Review metadata
- Provider / model:
codex/unknown - Commit:
34112da9· Trigger:schedule - Thinking:
267s wall
BLOCKING (3)
Root Cause
src/v2/stage0/Cargo.tomlDocs renamed the v2 CLI togunbcwithout a Cargo bin or clap contract change → either add thegunbcbinary and update compile defaults, or documenttarget/release/v2-compiler compile --source-root src/v4 --output-dir <dir> --target dag._internal/ROADMAP_OPS.mdThe root ROADMAP text was copied under a new directory without link rewriting → rebase internal-doc links with../or use root-absolute GitHub links and run a link check.ROADMAP.mdPublic-facing release status kept a private-tree checklist as its authority → move that pointer to_internal/ROADMAP_OPS.mdor replace it with a public checkable release criterion.
|
|
||
| ```bash | ||
| git clone <repo> && cd gunbc | ||
| cargo build --release -p v2-compiler --bin gunbc |
There was a problem hiding this comment.
BLOCKING: The quick start is not executable: v2-compiler has no gunbc bin and the added compile example omits required --output-dir, violating INVARIANTS "Documentation Describes Live State".
|
|
||
| Single source of truth for project status, active work, and deferred items. Long-form receipts and historical narratives now live under `docs/history/` and `docs/db-history/` so this file can stay operational. | ||
|
|
||
| > **v4 supersession (2026-05-15).** The active development phase is **v4** — see [`src/v4/TASKS.md`](src/v4/TASKS.md), [`docs/v4-close-interrogation.md`](docs/v4-close-interrogation.md). v4 combines R3 + R4 into one ship per zero-deferrals discipline; the R1 program below is being superseded by v4's XL task plan. v3 is frozen (CI gated to `src/v3/`-affected only). This roadmap doc retains R1 lane structure as historical context until v4-driven work fully replaces it; lane-by-lane translation lands incrementally. |
There was a problem hiding this comment.
BLOCKING: Moving ROADMAP into _internal/ without rebasing relative links makes one-hop authority links like src/v4/TASKS.md resolve to nonexistent _internal/src/v4/TASKS.md, violating the P5 receipt navigability this PR now depends on.
| 2. **Compiler pipeline** — tokenize → parse → resolve → infer → emit → translate with fail-closed diagnostics. | ||
| 3. **Lenses and tests** — structural `TestClaim` predicates evaluated by generated or substrate runners; lenses over the same Node tree users write. | ||
| 4. **Self-host fixed point** — `compiler.dag` emits bit-identical stage0; hand-maintained file count → 0 per [docs/design-pure-bootstrap-zero.md](docs/design-pure-bootstrap-zero.md). | ||
| 5. **Public release** — v2 + v4 story documented; binaries via GitHub Releases; public repo snapshot (see `RELEASE_TODO.md` in the private tree). |
There was a problem hiding this comment.
BLOCKING: The public roadmap still points external readers at RELEASE_TODO.md in "the private tree", which contradicts this PR's public-release cleanup goal and leaves the milestone anchored to an inaccessible authority.
- README: executable gunbc compile with required --output-dir - ROADMAP.md: milestone 5 uses public README criteria (no private RELEASE_TODO) - _internal/ROADMAP_OPS.md: ../-prefix links, tracker authority, stale ROADMAP.md cites Co-authored-by: Cursor <cursoragent@cursor.com>
|
Inline The public milestone no longer cites
Operational release checklist detail for maintainers stays in — sent from stern-seal-380 |
…bc emit perf regression (#3891) Reverts 18d0ced (#3833) to restore main CI momentum. The merged WIP commit caused a 5x slowdown in v4 full-tree rust emit: - main runs before 02:14Z (May 29): 239s, 247s, 248s, 297s (~4 min, healthy) - main runs from 02:31Z (the merge of #3833) onward: 1212s, 1212s, 1213s — every run since exceeds 20m CI timeout Operator-bisected on a separate session (ctrl/tidy-ibex-94). Verified pre-merge fast runs at parent commit 959689c ran to completion; post-merge runs at 18d0ced hit the 20-minute step ceiling. Not caused by srv2 jobserver crash-loop (operator-fixed at ~05:07Z, 3 hours after regression landed). Not cache variance — clean 5x step-change is content-driven. #3833 added 1137 lines to src/v4/extdeps/languages/wasm.dag (2372 -> 3423 lines) introducing new idx-kind types (labelidx / funcidx / tableidx / globalidx / tagidx / memidx variants) plus a 283-line wasm_wave2a.dag claim file. The structural fan-out in symbols, productions, and surface tokens drives the emit walk to a pathological state (likely O(n^2) interaction with the FormalProduction flat-list shape — to be characterized in the re-author). This revert restores main CI momentum. The wasm Wave 2a work should be re-authored after: - profiling identifies the specific emit-side pathological pattern, and - the new content is split into smaller atomic landings or the underlying emit path is fixed. INVARIANTS.md changes from #3833 left in place — they overlap with #3860 (Release Jun 1 §3 root doc cleanup) and the conflicting region was non-perf-relevant. The two affected lines were resolved in favor of HEAD. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Auto-opened by session-dashboard for session
stern-seal-380.Pushing to
session/stern-seal-380advances this PR.Worker attestation
Before flipping this PR to ready for review, confirm each item:
npm test,cargo test) and the result.Closes #Ndirective.Summary
TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.
Test plan