Skip to content

fleet-converge: dissolve bootstrap emitter scaffold + byte-oracle witness - #7007

Merged
briansrls merged 3 commits into
mainfrom
session/valiant-deer-438
Jul 21, 2026
Merged

briansrls merged 3 commits into
mainfrom
session/valiant-deer-438

Conversation

@gunbai-bot

@gunbai-bot gunbai-bot Bot commented Jul 21, 2026 •

Copy link
Copy Markdown
Contributor

Summary

Close out shell→dag P2 (Track 1, EMITTER lane): the fresh-standup bootstrap fragment already routes through orch_emit_pipeline(intent, Bash) on fresh_standup_bootstrap_intent (#6573/#6585). This PR:

  • Dissolves fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger and bootstrap_fragment_script_carrier_scaffold to Terminal (emitter routing is live; thin-run argv scaffold remains open).
  • Adds witness_fresh_standup_bootstrap_committed_byte_oracle_holds: exact arm→srv3 seam check against committed .github/fleet-converge.sh (newline-delimited, perturb-red).
  • Updates apply/emit witnesses for the dissolved bootstrap scaffold.

No change to emitted bytes — drift gate stays green.

Test plan

  • gunbc run --source-root dag --source-root src/v2 --entry dag/test/claim/fleet_converge_emit_test.dag --function fleet_converge_emit_bootstrap_holds --claim-run → true
  • gunbc run --source-root dag --source-root src/v2 --entry dag/test/claim/fleet_converge_emit_test.dag --function fleet_converge_emit_holds --claim-run → true
  • gunbc run --source-root dag --source-root src/v2 --entry dag/test/claim/fleet_converge_apply_witness_test.dag --function fleet_converge_apply_holds --claim-run → true

@gunbai-bot gunbai-bot Bot changed the title shell->dag Phase-2 (Track 1, EMITTER lane not migration): fleet_converge fresh-standup bootstrap emitter — route the fresh-standup bootstrap fragment through the v2 bash emitter rows (Procedure/Let band landed #6566), BYTE-ORACLE the emitted bootstrap vs the committed .github/fleet-converge.sh, and fleet-converge: dissolve bootstrap emitter scaffold + byte-oracle witness Jul 21, 2026
@gunbai-bot
gunbai-bot Bot marked this pull request as ready for review July 21, 2026 16:57
@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

Verified review 40856 (composer-2.5 APPROVE) against current HEAD 4940ab8a:

  • Bootstrap scaffold dissolution (fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger, bootstrap_fragment_script_carrier_scaffold → Terminal) matches live orch_emit_pipeline(intent, Bash) routing on fresh_standup_bootstrap_intent(); steady-state scaffolds remain Scaffold as intended.
  • witness_fresh_standup_bootstrap_committed_byte_oracle_holds correctly checks the arm→srv3 newline seam (4940ab8 fix), perturb-red, and poison controls alongside the existing full drift gate — no code changes required from this review.
  • Merge readiness: 1/2 distinct approvals (cursor only so far); CI still queued. Waiting on second approval + green checks per merge policy.

— sent from valiant-deer-438

@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

Verified review 40861 (claude-opus-4-7 APPROVE) against HEAD 4940ab8a — no code changes required:

  • Both scaffold→Terminal flips (fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger, bootstrap_fragment_script_carrier_scaffold) name the live orch_emit_pipeline(intent,Bash) routing on fresh_standup_bootstrap_intent; receipt witnesses co-flip in fleet_converge_emit_test.dag and fleet_converge_apply_witness_test.dag.
  • witness_fresh_standup_bootstrap_committed_byte_oracle_holds is the discriminating byte oracle (seam match, "x" perturb-red, poison guard) over committed .github/fleet-converge.sh; thin_invocation_argv_carrier_scaffold correctly remains Scaffold.

Merge readiness update: 2/2 distinct approvals (claude + cursor), no REQUEST_CHANGES, MERGEABLE — CI build still queued; waiting on green checks before operator merge.

— sent from valiant-deer-438

@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

CI fix pushed in 36bc0fa: extdeps_external_authority_gate_passes failed because main's okabe_ito/solarized color modules used bespoke authority row names (okabe_ito_authority, solarized_authority) instead of the required extdeps_external_authority_anchor. Renamed both anchors and updated gunbc.design.catalog provenance to qualified module paths.

Verified locally:

  • corpus_live_clean_tree_holds → true
  • fleet_converge_emit_holds → true

— sent from valiant-deer-438

@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

Verified review 40924 (composer-2.5 APPROVE) on HEAD 36bc0fa — no code changes required:

  • Bootstrap scaffold dissolution + orch_emit_pipeline(intent,Bash) routing on fresh_standup_bootstrap_intent is consistent with witnesses (witness_fresh_standup_bootstrap_committed_byte_oracle_holds, full drift gate).
  • extdeps_external_authority_anchor renames in okabe_ito/solarized fix the CI gate failure from bespoke authority row names; catalog.dag provenance uses qualified module paths.

Review 40925 (claude-opus-4-7) also APPROVEs the same HEAD. Merge readiness: 2/2 distinct approvals, no REQUEST_CHANGES, MERGEABLE — CI build still pending on 36bc0fa.

— sent from valiant-deer-438

@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

Verified review 40925 (claude-opus-4-7 APPROVE) on HEAD 36bc0fa — no code changes required:

  • Scaffold → Terminal on fleet_converge_fresh_standup_bootstrap_emit_dissolution_trigger and bootstrap_fragment_script_carrier_scaffold matches live orch_emit_pipeline(intent,Bash) routing; witness co-flips in fleet_converge_emit_test.dag / fleet_converge_apply_witness_test.dag are correct.
  • witness_fresh_standup_bootstrap_committed_byte_oracle_holds byte-oracles arm→srv3 seam vs committed .github/fleet-converge.sh with perturb-red and poison controls.
  • extdeps_external_authority_anchor renames + qualified catalog.dag paths fix the extdeps live-corpus gate.

Merge readiness: 2/2 approvals (claude + cursor on 36bc0fa), no REQUEST_CHANGES, MERGEABLE — CI build still pending.

— sent from valiant-deer-438

briansrls and others added 3 commits July 21, 2026 19:16
The committed .github/fleet-converge.sh separates the emitted bootstrap
arm from the srv3 thin-run line with a newline; the oracle must match
that seam exactly.

Co-authored-by: Cursor <cursoragent@cursor.com>
@briansrls
briansrls force-pushed the session/valiant-deer-438 branch from 36bc0fa to 99f46f5 Compare July 21, 2026 19:17
@gunbai-bot

gunbai-bot Bot commented Jul 21, 2026

Copy link
Copy Markdown
Contributor Author

Rebased session/valiant-deer-438 onto current main (99f46f5). Conflict in dag/gunbc/design/catalog.dag was mechanical: main already landed the extdeps_external_authority_anchor rename via #6991, so commit 36bc0fa (duplicate color-authority fix) was skipped during rebase. PR diff is back to the four fleet-converge files only.

— sent from valiant-deer-438

@briansrls
briansrls merged commit d36515b into main Jul 21, 2026
3 checks passed
@briansrls
briansrls deleted the session/valiant-deer-438 branch July 21, 2026 21:23
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant