ORCH-026: task dependencies (B waits for A) + single-repo merge serialization #78

Merged

admin merged 9 commits from feature/ORCH-026-b-a into main 2026-06-08 19:47:58 +03:00

Conversation 0 Commits 9 Files Changed 38 +2745 -5

admin commented 2026-06-08 19:07:01 +03:00

Owner

Copy Link

Summary

Implements ORCH-026 (ADR-001), two conditional levels — both kill-switched, never-raise, no new stage and no change to STAGE_TRANSITIONS / the QG_CHECKS registry.

• Level A — merge/deploy serialization within one repo. Reuses the existing ORCH-043/065 merge-lease (no new mechanism). The only new logic is an unconditional pre-merge rebase: in check_branch_mergeable, under the held lease, auto_rebase_onto_main is called always when premerge_rebase_always (default True), not only when the branch is behind — a deterministic structural anti-phantom complementing ORCH-073. No-op on an up-to-date branch (rebase keeps HEAD, push --force-with-lease -> "Everything up-to-date", CI not triggered). Kill-switch off -> ORCH-043 behaviour 1:1. Serialization window "merge -> main-updated" stays per-repo (cross-repo parallelism preserved).
• Level B — declarative task dependencies ("B waits for A"). Additive job_deps(task_id, depends_on_task_id) table (CREATE ... IF NOT EXISTS, safe on the live DB); claim_next_job NOT EXISTS gate defers a job whose depends-on tasks aren't yet done without occupying a max_concurrency slot; inert on empty job_deps -> zero regression. New leaf src/task_deps.py (never-raise): is_task_ready (fail-open), DFS cycle detection + Blocked/alert, declare_dependency, ingest_plane_relations (the db source never hits the network on the hot path), snapshot. Telegram waiting-line, /queue task_deps block, reconciler skip + cycle backstop; reaper untouched. Intra-repo only (v1).

Flags: ORCH_PREMERGE_REBASE_ALWAYS (true), ORCH_TASK_DEPS_ENABLED (true), ORCH_TASK_DEPS_SOURCE (db).

Self-hosting safety

Never pushes to main; the only force op stays --force-with-lease on the task branch. Prod container not restarted. Non-self repos remain a no-op on empty deps/scope.

Docs (golden source, same PR)

docs/architecture/README.md, CLAUDE.md, .env.example, CHANGELOG.md, ADR adr-0015 + work-item ADR-001.

Test plan

• tests/test_orch026_*.py — 64 tests (TC-A01..A08, TC-B01..B08, TC-G01/G02) green
• Regression: pinned premerge_rebase_always=False in the ORCH-043 short-circuit tests
• Full suite: pytest tests/ -q -> 991 passed
• ruff check clean on all changed/new source + tests (remaining 10 findings pre-exist on main)

🤖 Generated with Claude Code

## Summary Implements ORCH-026 (ADR-001), two conditional levels — both kill-switched, never-raise, **no new stage** and **no change to `STAGE_TRANSITIONS` / the `QG_CHECKS` registry**. - **Level A — merge/deploy serialization within one repo.** Reuses the existing ORCH-043/065 merge-lease (no new mechanism). The only new logic is an **unconditional pre-merge rebase**: in `check_branch_mergeable`, under the held lease, `auto_rebase_onto_main` is called **always** when `premerge_rebase_always` (default `True`), not only when the branch is behind — a deterministic structural anti-phantom complementing ORCH-073. No-op on an up-to-date branch (rebase keeps HEAD, `push --force-with-lease` -> "Everything up-to-date", CI not triggered). Kill-switch off -> ORCH-043 behaviour 1:1. Serialization window "merge -> main-updated" stays per-repo (cross-repo parallelism preserved). - **Level B — declarative task dependencies ("B waits for A").** Additive `job_deps(task_id, depends_on_task_id)` table (`CREATE ... IF NOT EXISTS`, safe on the live DB); `claim_next_job` `NOT EXISTS` gate defers a job whose depends-on tasks aren't yet `done` **without** occupying a `max_concurrency` slot; inert on empty `job_deps` -> zero regression. New leaf `src/task_deps.py` (never-raise): `is_task_ready` (fail-open), DFS cycle detection + `Blocked`/alert, `declare_dependency`, `ingest_plane_relations` (the `db` source never hits the network on the hot path), `snapshot`. Telegram waiting-line, `/queue` `task_deps` block, reconciler skip + cycle backstop; reaper untouched. Intra-repo only (v1). Flags: `ORCH_PREMERGE_REBASE_ALWAYS` (true), `ORCH_TASK_DEPS_ENABLED` (true), `ORCH_TASK_DEPS_SOURCE` (db). ## Self-hosting safety Never pushes to `main`; the only force op stays `--force-with-lease` on the task branch. Prod container not restarted. Non-self repos remain a no-op on empty deps/scope. ## Docs (golden source, same PR) `docs/architecture/README.md`, `CLAUDE.md`, `.env.example`, `CHANGELOG.md`, ADR `adr-0015` + work-item `ADR-001`. ## Test plan - [x] `tests/test_orch026_*.py` — 64 tests (TC-A01..A08, TC-B01..B08, TC-G01/G02) green - [x] Regression: pinned `premerge_rebase_always=False` in the ORCH-043 short-circuit tests - [x] Full suite: `pytest tests/ -q` -> **991 passed** - [x] `ruff check` clean on all changed/new source + tests (remaining 10 findings pre-exist on `main`) 🤖 Generated with Claude Code

admin added 6 commits 2026-06-08 19:17:45 +03:00

docs: init ORCH-026 business request 520bcafa73

analyst(ET): auto-commit from analyst run_id=386 518d7d18c8

architect(ET): auto-commit from architect run_id=387 9019e12d98

feat(ORCH-026): task dependencies (B waits for A) + single-repo merge serialization ... a74379f657

Level A — merge/deploy serialization within one repo: reuse the existing
ORCH-043/065 merge-lease (no new mechanism); the only new logic is an
unconditional pre-merge rebase in check_branch_mergeable — under the held
lease, auto_rebase_onto_main is ALWAYS called when premerge_rebase_always
(default True), not just when the branch is behind. No-op on an up-to-date
branch (rebase keeps HEAD, force-with-lease -> "Everything up-to-date", CI
not triggered). Kill-switch off -> ORCH-043 behaviour 1:1.

Level B — declarative task dependencies: additive job_deps table
(CREATE ... IF NOT EXISTS, no live-DB migration); claim_next_job gate
(NOT EXISTS) defers a job whose depends-on tasks are not yet 'done' without
occupying a max_concurrency slot; inert on empty job_deps -> zero regression.
New leaf src/task_deps.py (never-raise): is_task_ready (fail-open), DFS cycle
detection + Blocked/alert, declare/ingest_plane_relations (db source never
hits the network on the hot path), snapshot. Telegram waiting-line, /queue
observability, reconciler skip + cycle backstop, reaper untouched.

Invariants unchanged: STAGE_TRANSITIONS, QG_CHECKS registry (dep gate is a
claim_next_job врезка, not a registered QG), DB schema of existing tables,
HTTP endpoints; non-self repos remain a no-op on empty deps/scope.

Flags: ORCH_PREMERGE_REBASE_ALWAYS, ORCH_TASK_DEPS_ENABLED, ORCH_TASK_DEPS_SOURCE.
Docs: docs/architecture/README.md, CLAUDE.md, .env.example, CHANGELOG.md,
adr-0015. Tests: tests/test_orch026_*.py (64 tests); full suite 991 green.

Refs: ORCH-026

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

reviewer(ET): auto-commit from reviewer run_id=389 5b80f8facb

tester(ET): auto-commit from tester run_id=390 9800dc89e3

admin force-pushed feature/ORCH-026-b-a from 30d9f8e5ab to 9800dc89e3 2026-06-08 19:17:45 +03:00 Compare

admin added 1 commit 2026-06-08 19:28:42 +03:00

deploy(ORCH-036): finalize FAILED for ORCH-026 1d928dab57

admin added 1 commit 2026-06-08 19:33:58 +03:00

deploy(ORCH-036): finalize FAILED for ORCH-026 90a5cae8e6

admin merged commit b0e517c76a into main 2026-06-08 19:47:58 +03:00

admin referenced this issue from a commit 2026-06-08 19:47:59 +03:00

Merge pull request 'ORCH-026: task dependencies (B waits for A) + single-repo merge serialization' (#78) from feature/ORCH-026-b-a into main

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

No items

No Label

Milestone

No items

No Milestone

Projects

Clear projects

No project

Assignees

Clear assignees

admin

No Assignees

1 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: admin/orchestrator#78