2-compile-clean-shard-b — floor consumes the shards (TOP CI priority: dsl_compile_clean_gate is ~47min of the ~50min floor, width-immune). Shard A (#6118) landed partition+algebra+exemplar; B enrolls shard RUNNABLES in ci_floor_plan batches. Acceptance (8): no single serial compile-clean dominator; by briansrls · Pull Request #6127 · gunb-ai/gunbc

briansrls · 2026-07-02T04:07:55Z

Summary

Shard B wires compile-clean sharding into the CI floor plan: batch-1 runs a parallel RunnableCompileCleanShardBatch over the live module roster (replacing the serial dsl_compile_clean_gate monolith), with fail-closed verdict composition, coverage witnesses, and GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE rollback to the whole-tree gate.

Builds on Shard A (#6118): partition boundary, verdict algebra, exemplar transport.

Review fixes (`76a2d890f`)

Plan-coverage witnesses (non-tautological): witness_floor_shard_roster_coverage_in_plan and witness_floor_dropped_shard_red_control now destructure batch-0 RunnableCompileCleanShardBatch.entry_paths from gunbc_ci_floor_batches() via floor_batch_zero_compile_clean_shard_entry_paths() and compare against the roster — not compile_clean_shard_entry_paths() twice. gunbc_ci_floor_compile_clean_shard_count() likewise reads the plan, not the roster.
Rollback disposition: compile_clean_whole_tree_rollback_disposition (Scaffold + compile_clean_whole_tree_rollback_dissolution_trigger) on compile_clean_sharding_enabled.

Wall-clock receipt (acceptance #6 — partial, CI in flight)

Metric	Before (main monolith)	After (sharded, this PR)
Batch-1 compile-clean wall	~47 min serial `dsl_compile_clean_gate` (width-immune single node)	pending — CI run on `76a2d890f` in progress; executor emits `[measurement] compile-clean shard batch: N module shard(s), width=W, wall Xms`
Shard count	1 (whole-tree)	1605 module shards (plan == roster, witnessed)
Spawn width	1 (monolith)	budget-tree derived (srv1 cap)
Total floor wall	~50 min	pending CI
Peak RSS	monolith whole-tree resolve	pending — `[measurement] floor peak RSS` + `[calibration] max-per-shard-peak-rss` lines from `claim_executor`

"CI useful again" claim deferred until sharded batch-1 wall-clock receipt lands green on CI.

Test plan

gunbc run --source-root dsl --source-root src/v2 --entry src/v2/test/claim/ci_floor_plan_witness_test.dag --function ci_floor_plan_witnesses --claim-run → true
gunbc run --source-root dsl --source-root src/v2 --entry dsl/test/claim/dsl_compile_clean_shard_b_witness_test.dag --function compile_clean_shard_b_floor_plan_enrollment --claim-run → true
gunbc run --source-root dsl --source-root src/v2 --entry dsl/test/claim/dsl_compile_clean_shard_b_witness_test.dag --function compile_clean_shard_b_roster_and_coverage_red --claim-run → true
claim_executor --source-root dsl --source-root src/v2 --plan-entry src/v2/workflow/ci_floor_plan.dag --plan-function gunbc_ci_floor_batches — plan eval passes hermetic mode (shell.Env.Get mock for rollback env)
CI floor green on 76a2d890f

…rity: ds

gunbai-bot · 2026-07-02T06:18:53Z

Addressed both blocking review items in 76a2d890f:

1. Tautological plan-coverage witnesses — Valid finding. witness_floor_shard_roster_coverage_in_plan / witness_floor_dropped_shard_red_control were comparing the roster to itself. Fixed by extracting batch-0 RunnableCompileCleanShardBatch.entry_paths from gunbc_ci_floor_batches() (floor_batch_zero_compile_clean_shard_entry_paths → schedule_batch_compile_clean_shard_entry_paths) and comparing that to compile_clean_shard_entry_paths(). gunbc_ci_floor_compile_clean_shard_count() now reads the plan shard count (1605), not the roster helper — so witness_plan_shard_count_matches_roster is also non-tautological.

2. Rollback scaffold disposition — Valid. Added compile_clean_whole_tree_rollback_disposition (Scaffold bound to compile_clean_sharding_enabled) plus compile_clean_whole_tree_rollback_dissolution_trigger; witnessed in compile_clean_whole_tree_rollback_env_is_named.

PR body updated with the wall-clock receipt table (shard count + measurement hooks cited; full sharded batch-1 wall pending CI on this commit).

— sent from bright-ram-30

…rity: ds

…atch When batch-0 leads with RunnableCompileCleanShardBatch, plant a broken staged module entry instead of requiring a SingleClaim gating node. Preserves monolith-path SingleClaim planting for rollback plans. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-02T06:35:15Z

composer-2.5 --perturb-check finding — Valid (non-blocking in review, fixed anyway in latest commit).

With sharding on, batch-0's first runnable is CompileCleanShardBatch, so the old path that required a plantable SingleClaim gating node exited early. Fixed: --perturb-check now detects CompileCleanShardBatch as the gating plant target and writes a broken staged module (_perturb_compile_clean_shard_red.dag, same skew as existing perturb receipts) into the temp tree, then remaps the shard batch to compile only that entry. Monolith/rollback plans still use the existing SingleClaim plant path. Unit tests: perturb_plant_target_prefers_shard_batch_when_first, perturb_plant_target_uses_single_claim_for_monolith_batch.

— sent from bright-ram-30

gunbai-bot · 2026-07-02T06:37:02Z

claude-opus-4-7 per-thread resolve note — Valid observation, intentionally deferred (no commit).

run_compile_clean_shard_batch_node resolves COMPILE_CLEAN_SHARD_TRANSPORT once per worker thread because InterpContext is not Send (Rc internals — same reason run_memo_shared_claims documents "Runs on the main thread so InterpContext never crosses thread boundaries"). Parallelism here is at the shell-compile layer (run_shard_compile → shell.Exec.Run), not shared eval-context reuse. Width-folded duplicate resolves are real cost but bounded by spawn_width_cap; memoizing a Send-safe transport handle across threads is follow-on work, not a Shard B blocker.

— sent from bright-ram-30

…rity: ds

--perturb-check no longer hard-codes PERTURB_SHARD_BROKEN_MODULE; it evaluates broken_shard_module_source from the resolved shard transport graph (alias of perturb_module_source in dsl_compile_clean_transport). Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-02T06:48:41Z

claude-opus-4-7 PERTURB_SHARD_BROKEN_MODULE duplication — Valid; fixed in latest commit.

Removed the hand-Rust literal. --perturb-check now resolves dsl/tools/dsl_compile_clean_shard_transport.dag and evaluates broken_shard_module_source (.dag alias of perturb_module_source) via eval_data_item_value — same authority chain as witness_broken_staged_shard_compile_red. Unit test: perturb_shard_broken_module_source_reads_dag_authority.

— sent from bright-ram-30

gunbai-bot · 2026-07-02T07:01:02Z

composer-2.5 re-review — Verified: no findings, no code change required. Approval noted.

— sent from bright-ram-30

gunbai-bot · 2026-07-02T07:01:17Z

claude-opus-4-7 re-review — Verified: no findings, no code change required. Approval noted.

— sent from bright-ram-30

…rity: ds

Regenerate std_realization_schedule.rs from realization_schedule.dag (RunnableCompileCleanShardBatch, sharded schedule lens). Make run_compile_clean_shard_batch_node reject empty entry_paths instead of greening with zero work. Co-authored-by: Cursor <cursoragent@cursor.com>

gunbai-bot · 2026-07-02T12:33:08Z

composer-2.5 REQUEST_CHANGES — Both findings valid; addressed in 2e7fdb87a + latest commit.

Stage0 emit drift — Ran regen_stage0; std_realization_schedule.rs now includes RunnableCompileCleanShardBatch, compile_clean_sharded lens param, and CompileGateMonolithPresent / CompileCleanShardsInsufficient. regen_stage0 --verify green.
Empty shard batch fail-open — run_compile_clean_shard_batch_node now returns ok: false when entry_paths is empty (compile-clean-shard:empty entry_paths — fail-closed). Witness: compile_clean_shard_batch_empty_entry_paths_fails_closed.

— sent from bright-ram-30

gunbai-bot · 2026-07-02T12:40:56Z

composer-2.5 re-review (post-fix) — Verified: no findings, no code change required. APPROVE noted after e82f77f96 (regen + empty-shard fail-closed).

— sent from bright-ram-30

gunbai-bot · 2026-07-02T12:45:59Z

claude-opus-4-7 re-review — Verified both notes; no commit (explicitly non-blocking).

compile_clean_sharded bool vs plan introspection — Agree this is a parallel discriminator today: every call site derives it from compile_clean_sharding_enabled() (rollback env), which matches batch-0 shape while rollback exists. Dissolution trigger is already named on the rollback Scaffold; follow-up is to derive sharded-vs-monolith from batch-0 contents (compile_clean_shard_batch_label / RunnableCompileCleanShardBatch) and drop the third parameter when GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE retires — tracked, not merge-blocking.
path_param_value fold sentinel — Confirmed in regen output (std_http_path.rs); collateral from regen_stage0 sync, not shard-B logic. Noted for a separate regen/HTTP-path fix; out of scope here.

— sent from bright-ram-30

…rity: ds

gunbai-bot · 2026-07-02T14:30:57Z

composer-2.5 re-review (post env-fix) — Verified: no findings, no code change required. APPROVE noted on d02d57599.

— sent from bright-ram-30

gunbai-bot · 2026-07-02T14:36:02Z

claude-opus-4-7 re-review (post env-fix) — Verified all notes; no commit (explicitly non-blocking APPROVE).

Lens compile_clean_sharded fork — Agree; monolith branch is Scaffold-guarded with named dissolution trigger; sharded branch is canonical path.
eval_plan Wet for rollback env — Confirmed at claim_executor.rs:1041. Deliberate scaffold trade: plan shape depends on process env only while GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE exists; dissolves with rollback retirement. Will not widen env-gated rollbacks without an outer wrapper if more levers land.
spawn_width_cap silent 0 fallback — Confirmed; clamped to width≥1 downstream, consistent with RunnableDiscoveryBatch decode.

— sent from bright-ram-30

…egression) #6127 (merged 15:08 today) sharded compile-clean into ~1,605 per-module gunbc-compile subprocesses, each re-resolving its full import closure (sum-of-closures ~30.9x the monolith CPU), scheduled at WIDTH 1 because the width fold priced each single-module shard against a stale whole-corpus 4.24GiB RSS sample vs the 8GiB cap. Live regression: 8h57m wall / 8h07m CPU per job on an 87%-idle 128-core box. Pre-regression monolith green runs were 28-96min. Sets the modeled rollback valve GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1 on the ci job (env), restoring the whole-tree monolith compile-clean. Authority = the matching v2.workflow valve; rationale on ci_compile_clean_rollback_dissolution_trigger; Scaffold disposition bound, dissolves with the v2 authority once the in-process shard resolve pool lands. ci.yml regenerated via generated_artifact_gate main_wet; drift gate PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

The GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE valve does NOT cleanly restore a green monolith floor: under the valve, gunbc_ci_floor_schedule_lens_holds() FAILs and batch-0 carries neither shards nor the monolith gate label (verified by execution with claim_batch --wet under GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1). The valve looks like a half-exercised scaffold from #6127. Flipping it and patching witnesses to match an unverified plan path would violate DESIGN §5 (ship only what runs green), so this PR keeps only the sound, approved part: - gunbc_ci_job_timeout_policy_minutes = 30 (workflow-layer; GitHub 360 default stays modeled) - disposition text reconciled: 30min is deliberately BELOW today's 28-96min floor — a fail-fast smoke alarm until the resolver single-thread fix lands (verification local meanwhile) The #6127 regression is better addressed by a clean revert of #6127 or a proper fix to the plan's rollback branch (load-bearing floor-plan work) — surfaced to the operator. ci.yml regenerated (timeout-only, no rollback env); drift gate PASS; both shard-b floor witnesses PASS at origin/main state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…led GitHub default) (#6154) * WIP: fable * WIP: fable * WIP: fable * WIP: fable * Roadmap: fold in operator directives + 2026-07-01 live findings - §2 posture: required CI back on the fleet (#6111 revert; Ubicloud detour bought no wall-clock — serial compile-clean wall dominates); milestones updated; fleet-return group becomes fleet-hardening (the return happened ungated, the gates are now owed). - New: 2-compile-clean-serial (the ~47min width-immune dominant cost + its three levers), 2-runner-shape (operator 1:3 core:GiB + swap slot contract), cap-deconflation corrected (width=9 was live by construction; the deferred item was only the concept rename), CD-transport reframed post-revert, G4 gains the dup-run + lingering in_progress evidence. - §3 audit: 3-audit-artifact-freshness (operator freshness-gate ask; the .gitignore landed-drift incident as the RED receipt, closes via the merge-admission freshness block) and 3-audit-gate-disagreement (compile-clean green while discovery red on the same tree — the extdeps.shell collision; one tree, one verdict). - §1 5-defork gains the live evidence line (the shell fork kept main red). Witnesses re-targeted in lockstep; verified by execution: main_wet regen, drift gate, roadmap_authority_witnesses, doc-graph orphans+dangling all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap: operator review + fabric dispatch design + stateless-frontend MVP Applies the operator's three directives in one lockstep batch: 1. Roadmap-review blockers: - Ubicloud tone: detour was the diagnostic + emergency valve, not a failure; stays the break-glass/burst option. - CD transport reframed as target-host proof (LocalShell valid only when runner identity proves host=srv1), not Ubicloud-specific. - v1-collapse dependency edges now REAL: roadmap_spawner readiness is all-parents (node_dep_done folds every edge, not the first), with a RED/GREEN multi-parent witness; 5-collapse-v1 gains edges from dissolve-patches, test-migration, seed-honesty, defork. - workload-class admission model + cap/admission de-conflation (three facts, one knob today) + sound host admission (Guaranteed vs Burst with Σ-accounting receipts) + live host read seam as first-class nodes; G1 stale "dormant on Ubicloud" conditional fixed (cursor review finding on #6110). - Floor group renamed: throughput — reduce work, then schedule it. - Shelf: privacy/isolation model + structural correctness walls. 2. Fabric dispatch design (operator-signed): core design rule prose (ProviderOffer adapter / RunShape→Allocation→Receipt / strict default / oversubscription earned by receipts), StrictLease Policy 0, CI shape labels (runs-on as a projection; runner availability is backpressure), ProviderOffer rows + dormant-Ubicloud design-break probe, BurstLease Policy 1; TERMINAL reworded to "required CI runs on the compute fabric"; converge re-land gains the narrow srv2 first-acceptance case; ordering prose ①–⑩. 3. Stateless frontend MVP on fabric: first non-CI product-shaped consumer (SiteArtifact/WebServiceShape/DomainRoute/ServiceAllocation, manual DNS, digest-proving receipt, milestones A/B), service allocation receipt under control plane, fabric-hostable rule prose, shelf HTML line scoped. Verified by execution: main_wet regen, drift gate, roadmap_authority witnesses (new fabric pins + updated interleaving neighbor), spawner witnesses (incl. new all-parents RED control), doc-graph orphans + dangling links — all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * WIP: fable * Roadmap dispatch utility + first-receipt splits + converge edge-direction fix Operator ask: a small utility returning the active/needed work items to dispatch from the roadmap (frontend/visualization seam, useful now). - gunbc.roadmap_spawner gains roadmap_dispatch_json (single authority for spawn semantics): {ready (full node rows), upcoming (node + unmet_parents), done}. node_json also emits parent_node_ids (all parents — folds in claude's non-blocking #6110 finding; single parent_node_id kept for bridge back-compat). json_escape now escapes newlines (latent invalid-JSON bug, witnessed). - dsl/tools/roadmap_dispatch.dag: `claim_batch --source-root dsl --entry dsl/tools/roadmap_dispatch.dag --function main --wet` writes target/roadmap-dispatch.json (schema roadmap-dispatch-file/v1, embeds the M0-M3 worker-brief template). - First-receipt splits per operator dispatch review: 1-resolver-pathology → -a (profile receipt only) / -b (fix one confirmed pathology, edge b←a); 2-live-read-runner-memory as the first slice of the read seam (seam gated on it); 2-privilege-model sized with RED/green acceptance. - Edge-direction fix (operator structural concern): 2-converge-reland now DEPENDS ON 2-live-read-seam + 2-privilege-model (was inverted: privilege as converge's child); umbrella g2-runner edge removed so it cannot gate; periodic-actuation ← converge kept. Verified by execution: spawner witnesses (new dispatch-partition + newline-escape RED controls), authority witnesses, main_wet regen, drift gate, doc-graph — PASS; utility run produces valid JSON with 11 ready / 4 gated, gates matching the signed ordering. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap: dispatch compile-clean shard A (first-receipt split) + mark owned Operator go-ahead post-merge of #6110/#6111/#6112. 2-compile-clean-shard-a (partition boundary + ONE shard + compose proof, no scheduler enrollment) dispatched as adhoc-3e95c046-279 and marked owned; 2-compile-clean-shard-b (floor-plan enrollment + before/after batch-1 wall receipt) gated on A. Verified: regen, drift gate, authority + spawner witnesses PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Roadmap: mark 3-audit-gate-inventory done (#6116 merged) #6116 (gate-inventory audit + pilot red-control) merged without flipping its node, unlike #6120 which self-marked. Set done: true to parity with the resolver node — both now render "⏳ awaiting sign-off" (the [x] checkbox is gated on operator sign-off by design, not merge alone). Verified: main_wet regen, drift gate, roadmap_authority_witnesses PASS; dispatch file moves 3-audit-gate-inventory to the done partition. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Roadmap #6122: drop #6120 duplicate resolver node + heal doc-graph orphan Addresses cursor REQUEST_CHANGES on #6122: - Finding 3 (real): removed 1-resolver-pathology-receipt, a #6120-introduced near-verbatim duplicate of 1-resolver-pathology-a (§2 parallel representation). - Findings 1/2 (drift-heal, not regression): done:true renders "[ ] — ⏳ awaiting sign-off" by design (box = signoff_accepts, not merge); #6120 had committed ROADMAP.md with resolver-A as [x] with no sign-off — that was drift, and the regen heals it to the honest awaiting-sign-off state. - Also heals a pre-existing doc-graph orphan: mechanism-inventory-red-controls.md (#6116) was link-unreachable; attached it to the 3-audit-gate-inventory node via a new with_plan helper (keeps the node sized). Verified: main_wet, drift gate, roadmap_authority_witnesses, doc-graph orphans + dangling links all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap #6122: keep PR scoped — move hostname-converge demo off this branch Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Roadmap #6122: sync main (#6113/#6115/#6118) + mark those 3 merged nodes done The three verification-target PRs merged without self-marking their nodes; central-marker flips 2-cd-transport, 2-live-read-runner-memory, and 2-compile-clean-shard-a to done: true (render 'awaiting sign-off' — operator sign-off stays separate). Drift gate, roadmap witnesses, doc-graph all green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Roadmap: derive 2-debash-orchestration (#6106) + 5-dual-rep-lens (#6104) done; flip 2-privilege-model (#6114); mark 2-emit-partition dispatched - authored_merged_prs += 6104, 6106 (operator merged both) - 2-privilege-model done: #6114 merged, live receipts on #6123/#6130 wet runs - 2-emit-partition now un-gated and dispatched (adhoc-2040cdfe-46b) - regen ROADMAP.md via generated_artifact_gate main_wet; drift gate + roadmap witnesses PASS Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Revert throwaway deploy-principal edits (restore live_deploy/apply + ci_deploy_access to main) The manual-deploy investigation surfaced two real bugs (dash/pipefail + principal fiction) but the fix belongs in the modeled grounded-principal lane, not anemic stopgap strings. Restoring these two files to origin/main clean. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * CI policy: 30-minute job timeout (workflow layer; GitHub 360min default stays modeled) Operator directive 2026-07-02: a healthy floor is ~9min solo — a job past 30min is pathological and must fail fast + release its runner slot instead of squatting toward the platform ceiling (observed: 6h-timeout jobs holding all 25 fleet slots in a self-sustaining queue deadlock, ~9h single-core claim_executor runs). - gunbc_ci_job_timeout_policy_minutes: Int = 30 (+ Terminal disposition carrying the rationale) - extdeps default_job_timeout_minutes = 360 deliberately unchanged (§3: it faithfully models GitHub's real platform default; policy lives in the workflow layer) - ci.yml regenerated via generated_artifact_gate main_wet; drift gate PASS Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * CI emergency: roll back #6127 width-1 compile-clean sharding (30.9x regression) #6127 (merged 15:08 today) sharded compile-clean into ~1,605 per-module gunbc-compile subprocesses, each re-resolving its full import closure (sum-of-closures ~30.9x the monolith CPU), scheduled at WIDTH 1 because the width fold priced each single-module shard against a stale whole-corpus 4.24GiB RSS sample vs the 8GiB cap. Live regression: 8h57m wall / 8h07m CPU per job on an 87%-idle 128-core box. Pre-regression monolith green runs were 28-96min. Sets the modeled rollback valve GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1 on the ci job (env), restoring the whole-tree monolith compile-clean. Authority = the matching v2.workflow valve; rationale on ci_compile_clean_rollback_dissolution_trigger; Scaffold disposition bound, dissolves with the v2 authority once the in-process shard resolve pool lands. ci.yml regenerated via generated_artifact_gate main_wet; drift gate PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Back out #6127 rollback valve-flip; keep only the 30min timeout policy The GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE valve does NOT cleanly restore a green monolith floor: under the valve, gunbc_ci_floor_schedule_lens_holds() FAILs and batch-0 carries neither shards nor the monolith gate label (verified by execution with claim_batch --wet under GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1). The valve looks like a half-exercised scaffold from #6127. Flipping it and patching witnesses to match an unverified plan path would violate DESIGN §5 (ship only what runs green), so this PR keeps only the sound, approved part: - gunbc_ci_job_timeout_policy_minutes = 30 (workflow-layer; GitHub 360 default stays modeled) - disposition text reconciled: 30min is deliberately BELOW today's 28-96min floor — a fail-fast smoke alarm until the resolver single-thread fix lands (verification local meanwhile) The #6127 regression is better addressed by a clean revert of #6127 or a proper fix to the plan's rollback branch (load-bearing floor-plan work) — surfaced to the operator. ci.yml regenerated (timeout-only, no rollback env); drift gate PASS; both shard-b floor witnesses PASS at origin/main state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

…riority: dsl_compile_clean_gate is ~47min of the ~50min floor, width-immune). Shard A (#6118) landed partition+algebra+exemplar; B enrolls shard RUNNABLES in ci_floor_plan batches. Acceptance (8): no single serial compile-clean dominator; (#6127)" (#6156) This reverts commit 7541a5e. Co-authored-by: Brian Searls <briansrls@gunb.ai>

…#6164) * WIP: fable * WIP: fable * WIP: fable * WIP: fable * Roadmap: fold in operator directives + 2026-07-01 live findings - §2 posture: required CI back on the fleet (#6111 revert; Ubicloud detour bought no wall-clock — serial compile-clean wall dominates); milestones updated; fleet-return group becomes fleet-hardening (the return happened ungated, the gates are now owed). - New: 2-compile-clean-serial (the ~47min width-immune dominant cost + its three levers), 2-runner-shape (operator 1:3 core:GiB + swap slot contract), cap-deconflation corrected (width=9 was live by construction; the deferred item was only the concept rename), CD-transport reframed post-revert, G4 gains the dup-run + lingering in_progress evidence. - §3 audit: 3-audit-artifact-freshness (operator freshness-gate ask; the .gitignore landed-drift incident as the RED receipt, closes via the merge-admission freshness block) and 3-audit-gate-disagreement (compile-clean green while discovery red on the same tree — the extdeps.shell collision; one tree, one verdict). - §1 5-defork gains the live evidence line (the shell fork kept main red). Witnesses re-targeted in lockstep; verified by execution: main_wet regen, drift gate, roadmap_authority_witnesses, doc-graph orphans+dangling all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap: operator review + fabric dispatch design + stateless-frontend MVP Applies the operator's three directives in one lockstep batch: 1. Roadmap-review blockers: - Ubicloud tone: detour was the diagnostic + emergency valve, not a failure; stays the break-glass/burst option. - CD transport reframed as target-host proof (LocalShell valid only when runner identity proves host=srv1), not Ubicloud-specific. - v1-collapse dependency edges now REAL: roadmap_spawner readiness is all-parents (node_dep_done folds every edge, not the first), with a RED/GREEN multi-parent witness; 5-collapse-v1 gains edges from dissolve-patches, test-migration, seed-honesty, defork. - workload-class admission model + cap/admission de-conflation (three facts, one knob today) + sound host admission (Guaranteed vs Burst with Σ-accounting receipts) + live host read seam as first-class nodes; G1 stale "dormant on Ubicloud" conditional fixed (cursor review finding on #6110). - Floor group renamed: throughput — reduce work, then schedule it. - Shelf: privacy/isolation model + structural correctness walls. 2. Fabric dispatch design (operator-signed): core design rule prose (ProviderOffer adapter / RunShape→Allocation→Receipt / strict default / oversubscription earned by receipts), StrictLease Policy 0, CI shape labels (runs-on as a projection; runner availability is backpressure), ProviderOffer rows + dormant-Ubicloud design-break probe, BurstLease Policy 1; TERMINAL reworded to "required CI runs on the compute fabric"; converge re-land gains the narrow srv2 first-acceptance case; ordering prose ①–⑩. 3. Stateless frontend MVP on fabric: first non-CI product-shaped consumer (SiteArtifact/WebServiceShape/DomainRoute/ServiceAllocation, manual DNS, digest-proving receipt, milestones A/B), service allocation receipt under control plane, fabric-hostable rule prose, shelf HTML line scoped. Verified by execution: main_wet regen, drift gate, roadmap_authority witnesses (new fabric pins + updated interleaving neighbor), spawner witnesses (incl. new all-parents RED control), doc-graph orphans + dangling links — all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * WIP: fable * Roadmap dispatch utility + first-receipt splits + converge edge-direction fix Operator ask: a small utility returning the active/needed work items to dispatch from the roadmap (frontend/visualization seam, useful now). - gunbc.roadmap_spawner gains roadmap_dispatch_json (single authority for spawn semantics): {ready (full node rows), upcoming (node + unmet_parents), done}. node_json also emits parent_node_ids (all parents — folds in claude's non-blocking #6110 finding; single parent_node_id kept for bridge back-compat). json_escape now escapes newlines (latent invalid-JSON bug, witnessed). - dsl/tools/roadmap_dispatch.dag: `claim_batch --source-root dsl --entry dsl/tools/roadmap_dispatch.dag --function main --wet` writes target/roadmap-dispatch.json (schema roadmap-dispatch-file/v1, embeds the M0-M3 worker-brief template). - First-receipt splits per operator dispatch review: 1-resolver-pathology → -a (profile receipt only) / -b (fix one confirmed pathology, edge b←a); 2-live-read-runner-memory as the first slice of the read seam (seam gated on it); 2-privilege-model sized with RED/green acceptance. - Edge-direction fix (operator structural concern): 2-converge-reland now DEPENDS ON 2-live-read-seam + 2-privilege-model (was inverted: privilege as converge's child); umbrella g2-runner edge removed so it cannot gate; periodic-actuation ← converge kept. Verified by execution: spawner witnesses (new dispatch-partition + newline-escape RED controls), authority witnesses, main_wet regen, drift gate, doc-graph — PASS; utility run produces valid JSON with 11 ready / 4 gated, gates matching the signed ordering. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap: dispatch compile-clean shard A (first-receipt split) + mark owned Operator go-ahead post-merge of #6110/#6111/#6112. 2-compile-clean-shard-a (partition boundary + ONE shard + compose proof, no scheduler enrollment) dispatched as adhoc-3e95c046-279 and marked owned; 2-compile-clean-shard-b (floor-plan enrollment + before/after batch-1 wall receipt) gated on A. Verified: regen, drift gate, authority + spawner witnesses PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Roadmap: mark 3-audit-gate-inventory done (#6116 merged) #6116 (gate-inventory audit + pilot red-control) merged without flipping its node, unlike #6120 which self-marked. Set done: true to parity with the resolver node — both now render "⏳ awaiting sign-off" (the [x] checkbox is gated on operator sign-off by design, not merge alone). Verified: main_wet regen, drift gate, roadmap_authority_witnesses PASS; dispatch file moves 3-audit-gate-inventory to the done partition. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Roadmap #6122: drop #6120 duplicate resolver node + heal doc-graph orphan Addresses cursor REQUEST_CHANGES on #6122: - Finding 3 (real): removed 1-resolver-pathology-receipt, a #6120-introduced near-verbatim duplicate of 1-resolver-pathology-a (§2 parallel representation). - Findings 1/2 (drift-heal, not regression): done:true renders "[ ] — ⏳ awaiting sign-off" by design (box = signoff_accepts, not merge); #6120 had committed ROADMAP.md with resolver-A as [x] with no sign-off — that was drift, and the regen heals it to the honest awaiting-sign-off state. - Also heals a pre-existing doc-graph orphan: mechanism-inventory-red-controls.md (#6116) was link-unreachable; attached it to the 3-audit-gate-inventory node via a new with_plan helper (keeps the node sized). Verified: main_wet, drift gate, roadmap_authority_witnesses, doc-graph orphans + dangling links all PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * WIP: fable * Roadmap #6122: keep PR scoped — move hostname-converge demo off this branch Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * Roadmap #6122: sync main (#6113/#6115/#6118) + mark those 3 merged nodes done The three verification-target PRs merged without self-marking their nodes; central-marker flips 2-cd-transport, 2-live-read-runner-memory, and 2-compile-clean-shard-a to done: true (render 'awaiting sign-off' — operator sign-off stays separate). Drift gate, roadmap witnesses, doc-graph all green. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Roadmap: derive 2-debash-orchestration (#6106) + 5-dual-rep-lens (#6104) done; flip 2-privilege-model (#6114); mark 2-emit-partition dispatched - authored_merged_prs += 6104, 6106 (operator merged both) - 2-privilege-model done: #6114 merged, live receipts on #6123/#6130 wet runs - 2-emit-partition now un-gated and dispatched (adhoc-2040cdfe-46b) - regen ROADMAP.md via generated_artifact_gate main_wet; drift gate + roadmap witnesses PASS Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Revert throwaway deploy-principal edits (restore live_deploy/apply + ci_deploy_access to main) The manual-deploy investigation surfaced two real bugs (dash/pipefail + principal fiction) but the fix belongs in the modeled grounded-principal lane, not anemic stopgap strings. Restoring these two files to origin/main clean. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * CI policy: 30-minute job timeout (workflow layer; GitHub 360min default stays modeled) Operator directive 2026-07-02: a healthy floor is ~9min solo — a job past 30min is pathological and must fail fast + release its runner slot instead of squatting toward the platform ceiling (observed: 6h-timeout jobs holding all 25 fleet slots in a self-sustaining queue deadlock, ~9h single-core claim_executor runs). - gunbc_ci_job_timeout_policy_minutes: Int = 30 (+ Terminal disposition carrying the rationale) - extdeps default_job_timeout_minutes = 360 deliberately unchanged (§3: it faithfully models GitHub's real platform default; policy lives in the workflow layer) - ci.yml regenerated via generated_artifact_gate main_wet; drift gate PASS Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * CI emergency: roll back #6127 width-1 compile-clean sharding (30.9x regression) #6127 (merged 15:08 today) sharded compile-clean into ~1,605 per-module gunbc-compile subprocesses, each re-resolving its full import closure (sum-of-closures ~30.9x the monolith CPU), scheduled at WIDTH 1 because the width fold priced each single-module shard against a stale whole-corpus 4.24GiB RSS sample vs the 8GiB cap. Live regression: 8h57m wall / 8h07m CPU per job on an 87%-idle 128-core box. Pre-regression monolith green runs were 28-96min. Sets the modeled rollback valve GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1 on the ci job (env), restoring the whole-tree monolith compile-clean. Authority = the matching v2.workflow valve; rationale on ci_compile_clean_rollback_dissolution_trigger; Scaffold disposition bound, dissolves with the v2 authority once the in-process shard resolve pool lands. ci.yml regenerated via generated_artifact_gate main_wet; drift gate PASS. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * Back out #6127 rollback valve-flip; keep only the 30min timeout policy The GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE valve does NOT cleanly restore a green monolith floor: under the valve, gunbc_ci_floor_schedule_lens_holds() FAILs and batch-0 carries neither shards nor the monolith gate label (verified by execution with claim_batch --wet under GUNBC_CI_COMPILE_CLEAN_WHOLE_TREE=1). The valve looks like a half-exercised scaffold from #6127. Flipping it and patching witnesses to match an unverified plan path would violate DESIGN §5 (ship only what runs green), so this PR keeps only the sound, approved part: - gunbc_ci_job_timeout_policy_minutes = 30 (workflow-layer; GitHub 360 default stays modeled) - disposition text reconciled: 30min is deliberately BELOW today's 28-96min floor — a fail-fast smoke alarm until the resolver single-thread fix lands (verification local meanwhile) The #6127 regression is better addressed by a clean revert of #6127 or a proper fix to the plan's rollback branch (load-bearing floor-plan work) — surfaced to the operator. ci.yml regenerated (timeout-only, no rollback env); drift gate PASS; both shard-b floor witnesses PASS at origin/main state. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * docs: land enforcement-intent governance model (DESIGN thread + ROADMAP §3 node + sketch) Model the operator's recurring standing directives (complexity repo-wide, lenses-must-be-live, self-application, no dual-representation/anemia) as durable StandingIntent rows, gated fail-closed against LensContract + coverage receipts -- "ask once, compile forever". One new authority (StandingIntent); everything else extends existing machinery (LensRegistryEntryV0 -> LensContract; reuse ConstructionJustification / subject_roster; consume intent_linearity / self_applying_lenses for the fractal/self-application layer). Un-shelves the lens-meta-wall lane into ROADMAP §3 with displaced-cost justification (the operator's repeated manual what-is-enforced-by-what-lens-over-what-corpus join). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * docs: enforcement-intent review refinements (consumer/mode/receipt/exemption types + roadmap A-E split) Operator design-review deltas before the model becomes load-bearing: - ConsumerKind / ConsumerRequirement replace the hardwired FloorGate — merge-admission, pre-push, periodic-actuator, deploy-readback are valid consumers (gate leg 4 = consumer_satisfies). - EnforcementMode order (Advisory < AuditOnly < Blocking) modeled with enforcement_mode_satisfies; no overloaded >=. - CoverageReceipt shape added; the gate reads receipts, never self-declared contract claims (a red_control:Present whose receipt is RedControlFailedToFlip still reds). - ScopeSatisfaction / NarrowingReason type the whole-corpus-narrowing path (BootstrapBlocked / TypeReflectionUnavailable / ExternalRuntimeOnly / ExplicitOperatorExemption) — not a stringly hatch. - StandingIntent admission rule (recurring + displaced cost + scoped + mechanism class + receiptable) keeps the carrier from becoming a preference dump. - ROADMAP node split into parent + children A-E with explicit A->B->C-before-object-rules dependency. DESIGN.md entry unchanged (its "mechanism claiming enforcement" wording was already the refinement). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * WIP: fable * docs: fix #6164 review — edit .dag AUTHORITIES not generated artifacts; de-nickname PropertyClass Addresses cursor REQUEST_CHANGES (all three findings valid): - DESIGN.md / ROADMAP.md are EMITTED from gunbc.design_document / gunbc.roadmap_authority. My hand-edits to the generated artifacts were drift (a §3 parallel-representation of the authority, and would fail the generated-artifact drift gate). Moved the enforcement-intent content INTO the authorities (open_threads_blocks li + section_3 node + un-shelve prose) and regenerated (claim_batch main_wet). Drift gate now PASSES: committed == emitted. - Removed the PropertyClass nickname (git-grep empty; a fork of the existing LensIdV0 lens-property coproduct at src/v2/lens/registry.dag:10) — the exact §3 sin this PR argues against. StandingIntent.property now reuses LensIdV0. Fitting that a single-authority PR forked its own authority; caught by review, fixed at the source. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Brian Searls <briansrls@gunb.ai> Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

…solver-B #6155) (#6170) #6105 (runtime provenance ingest) was closed as not-landable: a floor-repair scout measured it as work-neutral theater — pruning INCREASES resolve+typecheck (PRESENT 6/28/3885ms vs run-all 1/27/2042ms) and only saves the witness EVAL, because the node-frontier skip fires AFTER the witness-entry closure is resolved and floor_runner resolves its closure twice. Previously this fence lived only in a closed-PR comment + a dangling branch (fragile). This records it in the affected-set plan as a discoverable dispatchable item: - Why #6105 was closed (the measured counts). - Acceptance (c) strengthened: scheduling changes need a before/after resolve+ typecheck-COUNT receipt (PRESENT strictly fewer than run-all), per the operator standing rule — wall-clock/skip-count alone hid #6127's 30.9x regression. - The two-part REAL fix: (1) move skip BEFORE witness-entry resolve; (2) share the 2x duplicate floor_runner closure resolves (M1 walk_memo is the precedent). - Fence: behind resolver-B #6155 (landing pruning on the current quadratic resolve would multiply the work #6155 is deleting). Branch session/quick-carp-343 preserved as harvest. Doc-only. Co-authored-by: Brian Searls <briansearls1@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Brian Searls added 6 commits July 2, 2026 04:07

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

c7c4692

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

9fa193c

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

3f68599

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

7b2fb2c

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

a03a5ba

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

16faeed

…rity: ds

gunbai-bot Bot marked this pull request as ready for review July 2, 2026 06:06

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

76a2d89

…rity: ds

Brian Searls and others added 2 commits July 2, 2026 06:33

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

5b4d09e

…rity: ds

Brian Searls and others added 2 commits July 2, 2026 06:46

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

bda5001

…rity: ds

Brian Searls and others added 5 commits July 2, 2026 07:06

Merge remote-tracking branch 'origin/main' into session/bright-ram-30

4b419ae

Merge remote-tracking branch 'origin/main' into session/bright-ram-30

2357b20

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

2d2da18

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

2e7fdb8

…rity: ds

WIP: 2-compile-clean-shard-b — floor consumes the shards (TOP CI prio…

d02d575

…rity: ds

briansrls merged commit 7541a5e into main Jul 2, 2026
1 of 2 checks passed

briansrls deleted the session/bright-ram-30 branch July 2, 2026 15:08

gunbai-bot Bot mentioned this pull request Jul 2, 2026

CI policy: 30-minute job timeout (workflow-layer policy over the modeled GitHub default) #6154

Merged

gunbai-bot Bot mentioned this pull request Jul 2, 2026

Revert #6127 (compile-clean shard-b): width-1 sharding was a 30.9x regression #6156

Merged

gunbai-bot Bot mentioned this pull request Jul 2, 2026

affected-set: record the fenced REAL-fix follow-on (behind resolver-B #6155) #6170

Merged

Navigation Menu

Repository navigation

briansrls commented Jul 2, 2026 •

edited by gunbai-bot Bot

Loading

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

briansrls commented Jul 2, 2026 • edited by gunbai-bot Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Review fixes (76a2d890f)

Wall-clock receipt (acceptance #6 — partial, CI in flight)

Test plan

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

gunbai-bot Bot commented Jul 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

briansrls commented Jul 2, 2026 •

edited by gunbai-bot Bot

Loading

Review fixes (`76a2d890f`)