[wangbill (YunchuWang)]

Summary

What changed?

• Add opt-in payload-aware client and worker options, unary request externalization and response hydration without changing the disabled transport path.
• Hydrate work items and history chunks before user execution/conversion, apply streamed backpressure, propagate storage failures as retryable work, and handle large histories without variadic argument limits.
• Await bounded, cancellation-aware abandonment of acquired work when payload processing fails without a converted business completion; tokenless host entrypoints preserve original errors.
• Scope completion-wait deadline retries to scheduler RPC failures, not successful-response payload hydration or custom-store deadlines.
• Centralize unary payload/metadata/RPC/hydration in _call. Scheduling, completion waits, rewind and restart supply only their RPC error-mapping/retry callbacks; storage/metadata/hydration errors retain their original identity.
• Expose Azure-managed builder opt-in with core >=0.4.1 and configuration-preservation tests.
• Use a custom in-memory PayloadStore in loopback gRPC tests. No Azure Blob provider, cleanup API, reserved task registration, or future purge import exists in this snapshot.

Why is this change needed?

• Phase 2/6 of the requested large-payload implementation split. This PR's base is phase 1's branch, so its diff excludes previously reviewed phases.
• Behavioral reference: durabletask-dotnet 91739687. The stack also inherits the reviewed JavaScript carryover and whole-message size-preflight repairs described below; it does not claim unchanged .NET failure-priority semantics.

Issues / work items

• Resolves: N/A (user-requested phased feature review).
• Related: original oversized PR #387. It is closed without merging; its original branch/history remain preserved.

---

Project checklist

• Release notes are not required for the next release
  • Otherwise: Notes added to CHANGELOG.md
• Backport is not required
  • Otherwise: Backport tracked by issue/PR #issue_or_pr
• All required tests have been added/updated (unit tests, E2E tests)
• Breaking change?
  • If yes:
    • Impact: The Azure-managed package now requires core >=0.4.1 for its public payload option types; core 0.4.0 is not a valid type/API provider.
    • Migration guidance: Publish the completed core 0.4.1 before the dependent package; do not release this partial phase.

---

AI-assisted code disclosure (required)

Was an AI tool used? (select one)

• No
• Yes, AI helped write parts of this PR (e.g., GitHub Copilot)
• Yes, an AI agent generated most of this PR

If AI was used:

• Tool(s): GitHub Copilot.
• AI-assisted areas/files: This phase's source, regression tests, and directly related metadata/documentation.
• What you changed after AI output: Split the implementation along independently buildable boundaries; retained focused regression evidence and removed unrelated formatting-only changes. Human review is still required.

AI verification (required if AI was used):

• I understand the code and can explain it
• I verified referenced APIs/types exist and are correct
• I reviewed edge cases/failure paths (timeouts, retries, cancellation, exceptions)
• I reviewed concurrency/async behavior
• I checked for unintended breaking or behavior changes

These human attestations are intentionally left unchecked.

---

Testing

Automated tests

• Current review head: f3ea69fc79eb594961ab4d57651f5e1360fb9127.
• npm run test:unit -w @microsoft/durabletask-js -- --runTestsByPath test\client-error-cause.spec.ts test\client-wait.spec.ts test\client-wait-grpc.spec.ts test\payload-transport.spec.ts test\orchestration-id-reuse-policy.spec.ts test\client-options.spec.ts test\client-suspend-resume.spec.ts test\client-purge.spec.ts — 233 tests / 8 suites passed on this exact approved commit.
• npm run build:core and npm run lint — passed. Scoped TypeScript/changelog formatting, commit-hook ESLint and git diff --check b4ddad9afa9f28d1ca9ebc510e5921074855cd56..HEAD passed.
• TDD reproduced 18 remaining rewind/restart upload/metadata identity failures before production edits. The final suite adds 56 unique controls, including frozen/reused errors, actual scheduler causes, cancellation/backoff stages and metadata failure on a later retry.
• The direct eligible-input control verifies one real in-memory upload across three RPC attempts, identical request/token, fresh metadata three times, two policy invocations and hydration only after success. Existing real-loopback scheduling/wait fixtures and old purge tests are unchanged; no sleep was added to hide default-dispatch changes.
• Owning receipts: central-call-red.log, central-call-owning-final-green.log, central-call-final-build-lint.log, central-call-final-format-diff.log; failed intermediate fixture/timing logs are not presented as passing final evidence.
• git diff --check origin/yunchuwang-payload-phase-1-core...HEAD — passed. Prior heads remain ancestors; predecessor bases, phase-local review scopes, release/cleanup/version notes and ready-for-review state are preserved.
• Combined final on exact ae63cab8ecb6ec6b2b3e0e15067091a5b8a6b24e: npm run test:unit — 3,180 tests / 133 suites passed (core 2,028; Functions 226; Blob 39; Azure-managed 821; export-history 66). Ten opt-in service tests are skipped by this unit command.
• On the same final head, npm run build passed all five workspaces and npm run lint passed. This is the true final aggregate containing both approved commits; the earlier stopped pre-lifecycle run is not a final pass.
• Final structural/blob audit passed: one unary transport site and one upload/resolve site in _call; all five semantic callers delegate; phase4–6 client/lifecycle fixtures match the approved runner fix, and PR1 processor plus existing public-loopback fixtures are unchanged.
• Final DTS live run: 39837123528 on exact ae63cab8ecb6ec6b2b3e0e15067091a5b8a6b24e — PENDING (in_progress) at 2026-10-22T19:07:38.5343134Z. The link is authoritative for later completion; no early pass is claimed.
• PR388 source/body remain unchanged at 137bb547e8781fd84c2738c3e13ae0f65cd66174; its separately verified hosted evidence is 30 SUCCESS / 1 SKIPPED. Review approval is still required. Main-base-only PR filters are unchanged; absent stacked-PR jobs are not passes.
• Scope limits remain: no live storage-outage or mixed .NET/JavaScript execution, no transactional rollback or generic orphan collector. Tombstone cleanup cannot discover blobs without a backend-recorded reference.

Manual validation (only if runtime/behavior changed)

• Environment (OS, Node.js version, components): Windows, Node.js v24.14.0; loopback gRPC transport tests and provider-independent in-memory payload storage.
• Steps + observed results:
  1. Materialized this phase on its declared predecessor, with no future-phase imports.
  2. Ran the exact owning checks above and inspected the incremental patch.
  3. Kept the original feature branch/history and primary user checkout unchanged.
• Evidence (optional): Full local logs are session artifacts, not repository changes.

---

Notes for reviewers

Shared unary boundary correction

• f3ea69fc79eb594961ab4d57651f5e1360fb9127 restores one private _call pipeline: combine cancellation once; externalize once; generate fresh metadata outside each RPC catch; invoke only the operation's RPC policy; hydrate once after success. true retries the RPC, false rethrows unchanged, and throwing maps a genuine RPC error with its cause. The no-metadata path preserves its original dispatch timing.
• Scheduling and completion waits no longer inline separate pipelines; rewind/restart mappings use the same RPC-only boundary. Errors are not mutated or tagged and no public API, helper signature, dependency or middleware framework was added.
• The separate auto-purge runner dedupe follow-up belongs to #391, not this phase.
• All earlier reviewed carryover, UTF-8 preflight, token/cancellation, rewind mapping and worker-release repairs remain ancestors. No PR merge, package publication or history rewrite was performed.

Merge order	Phase	PR	Base branch
1	Core contracts	#388	main
2	SDK integration	#389	yunchuwang-payload-phase-1-core
3	Azure Blob provider	#390	yunchuwang-payload-phase-2-sdk
4	Purge protocol/client	#391	yunchuwang-payload-phase-3-blob
5	Purge worker	#392	yunchuwang-payload-phase-4-client
6	Release/usage/consumers	#393	yunchuwang-payload-phase-5-worker

Merge 1 → 2 → 3 → 4 → 5 → 6. Each PR targets its predecessor; retarget the next PR only after its prerequisite merges. Do not publish packages or describe a partial phase as the complete feature.

• This phase enables transparent transport using a supplied PayloadStore, but does not provide Azure Blob storage or cleanup.
• Payload-aware work-item capability is advertised only when enabled. Cleanup filters and business-version exemptions are deliberately deferred to phase 5.
• Full live DTS/mixed-.NET execution has not been run locally (Docker daemon unavailable). The final phase records aggregate local Blob/Azurite, source/wire interoperability and packed-consumer evidence without claiming absent hosted checks.

[wangbill (YunchuWang)]

what no max retries?