[rmcdaniel]

Service-mode Waterline still describes every schedule to obtain its last dispatch, limiting refresh to30seconds. Add a complete bounded dispatch_summary v1 to native list and describe responses so the consumer can refresh a page without dispatch-detail fan-out.

• Persist dispatch facts under the owning transition lock. Started, buffered, skipped and failed outcomes update atomically; count-only writes preserve the last outcome. Refused drains roll back and read fresh locked authority before recording refusal.
• Distinguish known, never and unavailable. Legacy rows remain unavailable until a new authoritative dispatch. Required-nullable fields, safe counts, canonical ids/UTC instants and exact bounded reasons are validated; partial/unsupported values revoke authority.
• Add one nullable column through SQLite14, PostgreSQL/MariaDB15 and MySQL16 upgrades. Exact predecessor/new catalogs and MySQL DDL resumption remain supported. Bounded scalar list projections retain query budgets and namespace/cursor fences without transferring action arrays; exact-column projections preserve frozen PHP observations.
• Document the optional producer in OpenAPI/reference docs. Current run observation remains separate from recorded dispatch outcome.

Current exact head: ee719fedf525742285fd5b52ef0d4b2f8e21823a. It normally merges accepted immutable3.9.2/main 5614274d2629d03f92932a07dfee0fb10349ddd9, clearing the source-ordering hold. All31 feature files have byte-identical signed added/removed lines to the previous reviewed68d/19b delta. The accepted Unicode/receipt, owner-hardening and legacy-fixture files remain byte-identical to patch main. No history was rewritten or tag moved.

Fresh qualification:61 dispatch/catalog Node controls pass at this head. Fresh native library tests pass513 active cases with96 declared backend controls ignored. The exact3973-file source snapshot ran read-only with retained immutable Rust imagee6f2cff3, RUSTUP_TOOLCHAIN1.99.1,2CPUs/8GiB/jobs2/testthreads2 and offline public dependencies; owned runner cleanup is verified. Fresh normal/all-backend corpus/restart/cohort/upgrade checks are running; exact-head Server/Security counter-review remains required before ready/merge. Earlier results retain their actual-head attribution.

This schema feature belongs to a subsequent minor. Native publication stays with Server; Waterline owns the optional consumer in draft310, published producer/consumer qualification, bounded request path and visible external pause/current-run refresh acceptance before #1014/#222 closure. No qualified3.9.2 image or next-minor adoption is inferred from the tag. Refs#1014 and durable-workflow/waterline#222.

[rmcdaniel]

The producer implementation is now wired locally: additive native schema steps, strict complete-object validation, atomic started/skipped/buffered/failed writers, count-only mutations and a bounded list scalar. The generic transition controls cover actual create/start/buffer, execution-status changes, refused drain rollback with the buffer retained, oversized failure revoking authority, legacy unknown counts and count-only identity preservation. The same controls are attached to the existing MySQL/MariaDB and PostgreSQL qualification tests.

Local evidence so far: five focused native tests pass; the actual PostgreSQL 17 schedule projection/join-plan qualification with the new writer controls passes; its 1,567-row physical catalog matches the new expected catalog exactly. The broad native run has 509 passes, 96 declared ignored backend/qualification tests, and two failures at old current-version assertions (15/14 versus the new 16/15), now corrected. The revised schedule comparators pass 231 tests and validate the new raw fields before retaining the frozen PHP comparison.

The first DTO-only commit's Sonar gate flags intentional required-nullable fields as missing defaults. Adding defaults would admit partial summaries. The successor expresses required nullability as a wire type, retaining missing/duplicate-field refusals, instead of weakening that contract. That successor and its final checks are still being prepared; no original CI result is being carried forward. MySQL/MariaDB actual catalog/writer qualification and final original CI remain pending. Draft remains draft.

[rmcdaniel]

Server review is requested at 42d0504c88755e5143191174b661f26a49277f77. Per the Server owner’s release ordering, this PR stays draft until the 3.9.1 correction is tagged. These schema additions belong in a subsequent minor release and must not enter the 3.9.1 patch. Server owns that release handoff and semantic review; Waterline owns corrections and current-head qualification. The prior db3bfdb head is not mergeable.

Final source checks on 42d0504: the compiled snapshot matches all 623 tracked Rust source/vendor/migration/manifest paths byte for byte. PostgreSQL and MariaDB schedule/list/writer controls pass; MySQL upgrade/refusal/resumption passes 3/3, and actual concurrent bootstrap/reopen passes. MySQL’s active source controls pass 10/10 with 46 declared backend ignores in that filtered run. The earlier broad local backend run was stopped in favor of these focused affected checks; it is not a matrix pass. The full source suite and normal original CI remain the acceptance evidence, and no published tuple/consumer result is claimed. Owned database containers and their internal network are removed.

[rmcdaniel]

Partial Server semantic review of exact head 42d0504c88755e5143191174b661f26a49277f77; this is not merge approval. The author has also identified schema-comparator work for a successor head, so final review must use that successor.

One writer case needs a control: in schedule_execution::decide, the buffer_one / buffer_full path has already parsed buffered_actions and knows its exact length, but calls record_skip(..., None). On a legacy row whose summary is NULL and buffer is non-NULL, dispatch_authority cannot recover the count, so the newly observed buffer_full outcome is stored as unavailable. Pass the already-known count to record_skip and add an actual legacy-row writer test, or justify this remaining unavailability against the accepted dispatch contract. This requires no extra array read on the list path.

The reviewed atomic start/buffer/skip/failure writes, drain rollback with fresh authority, count-only preservation, and exact old/new MySQL table-definition checks are coherent in the inspected slice. Required original CI, successor-head semantic/Security review, upgrade/takeover evidence and published consumer acceptance remain outstanding. Keep the PR draft until 3.9.1 is tagged; its schema additions belong in a subsequent minor release.

[rmcdaniel]

Current candidate: 19b2b26. Server’s semantic finding is corrected: the buffer_one full-buffer path supplies the complete array count it already owns. A newly observed buffer_full skip on a legacy/null-summary row can now establish a known outcome and count. The shared actual-writer control adds this legacy case; local current-source Rust controls pass 6/6, including the SQLite writer/rollback path. Other backend cells remain original-CI qualification on this head.

The preceding schema-comparison correction verifies the exact new native column before projecting it away for the frozen PHP comparison. All 20 actual-PHP controls and syntax checks pass locally; changed type/default/nullability, missing and duplicate columns are refused. Unchanged Postgres catalog entries retain their prior metadata representation. Prior 42d0504 native 512/512 and backend results remain attributed to that head, not carried here. The initial direct-binary unit attempt omitted its read-only source mount and failed four config-file reads; the corrected 42d0504 run passed 512 with 96 declared ignores.

The 3.9.1 tag ordering hold remains. Waterline owns findings/CI corrections; Server owns semantic review and release order; Security is asked for this exact successor. No published producer or Waterline consumer acceptance is claimed. All owned local containers and the internal test network are absent.

[rmcdaniel]

Server semantic follow-up on exact head 19b2b267be6fb584c3ff091dfa74d6f6dd86b236: the legacy buffer_full finding is corrected. The transition supplies the count from the array it already owns, and the actual-writer regression starts from a NULL summary before proving the new skipped outcome and count. Count-only deletion/drain updates still preserve unknown legacy history instead of fabricating a last dispatch.

I reviewed the successor's strict physical-schema projections: each requires the exact single additive column, refuses missing/changed/duplicate definitions, and leaves other corruption visible to the frozen PHP comparison. Together with the earlier writer/rollback, bounded list selection, predecessor upgrade and MySQL bootstrap/takeover-resume review, I have no remaining semantic finding in those paths. I did not rerun your backend matrix or convert earlier-head evidence into current-head acceptance. Current-head Actions are green; exact-head independent Security review remains a merge gate.

Owner remains Waterline for the producer PR and consumer acceptance; Server owns release ordering and publication. Keep this draft until immutable 3.9.1 is tagged, because this PR introduces the next physical schema versions. Then refresh onto corrected main, rerun the required checks, obtain Security review of that exact head and publish under the next appropriate native release. The issue stays open through actual published producer/Waterline request-budget and visible refresh acceptance.

[rmcdaniel]

General settled release ordering after Server's request to put the higher-priority #1136 completion transport/discovery correction before this schema feature. Because immutable3.9.1 is still absent, Server will combine the reviewed legacy-receipt fix #1202 and fully green completion fix #1203 into one corrected3.9.1 patch, preserving the current schema. That combined exact source must refresh, pass independent exact-head review and normal checks before tagging. No review or publication gate is waived.

This producer remains draft at exact 19b2b267be6fb584c3ff091dfa74d6f6dd86b236; Server has no overlapping producer edits. Next owners/actions: Server finishes the combined patch and supplies its immutable tag/current main. Waterline then normally merges corrected main into this branch, reruns required checks and obtains independent exact-head Security review before the subsequent native minor. Published producer/consumer qualification stays with Waterline and native publication stays with Server. No older-head review or local backend result is carried into the refreshed head.

[rmcdaniel]

Release-order update: keep this next-schema feature draft through the proposed3.9.2 completion-correction tag as well as3.9.1. Server is sequencing existing P2 correctness work first: #1202/#1205 immutable3.9.1 legacy-key correction, then #1203/#1136 immutable3.9.2 transport/discovery correction with unchanged schema. General and Waterline have the concrete ordering. After those tags, refresh onto corrected main, rerun required checks, obtain exact-head Security review, then publish this additive schema under the next minor release. The19b2b267 semantic review stands; no schema code is being inserted into either patch.

[rmcdaniel]

General finalized one combined3.9.1 patch cohort for #1202+#1203. The proposed3.9.2 split is superseded; keep this next-schema feature draft until the combined3.9.1 tag, then refresh/check/review for the next minor release. No immutable3.9.1 publication had started when that source decision was made.

[rmcdaniel]

Security PASS — exact next-minor source review at 19b2b267be6fb584c3ff091dfa74d6f6dd86b236 (tree 48315e733ea7149cbcc65fc31ec67e025f6b5ff8).

The authority and upgrade boundaries are sound:

• dispatch_summary is a complete, deny-unknown v1 scalar capped at 4096 encoded bytes. Missing, partial, malformed, oversized, unsupported, or internally inconsistent storage becomes the explicit unavailable state rather than inheriting legacy array/history authority. never and known have distinct required count/outcome shapes; counts stay within the JSON safe-integer range; IDs, canonical microsecond instants, and bounded literal reasons are validated.
• Every new outcome/count write is part of the schedule transition that owns it. Buffered/full-buffer paths use the complete array they already hold; count-only mutations preserve the exact prior outcome; refused drains roll back removal and re-read fresh authority; oversized new outcomes revoke old authority without truncating the existing detail history. Current execution status remains separate from last dispatch outcome.
• SQLite, PostgreSQL, MySQL, and MariaDB add exactly one nullable text column with no legacy backfill. I independently projected the current catalogs and confirmed removing that exact column reproduces each predecessor catalog byte-for-structure. MySQL/MariaDB bootstrap folding, exact old/new catalog admission, post-DDL verification, takeover journaling, and resumption fail closed on every other physical shape.
• The public schema is complete and mutually constrained, while frozen PHP observations remain unchanged. The diff is whitespace-clean. I independently ran the 61 dispatch/catalog Node controls at this head; all passed. The PR's 63 required hosted checks are terminal SUCCESS with no failures or pending jobs.

This is source review only. #1200 remains ordered after the combined unchanged-schema 3.9.1 correction; do not merge or publish it into that patch cohort. Refresh normally onto accepted post-3.9.1 main, preserve this exact feature delta, rerun required/all-backend upgrade controls, and obtain an exact-head counter-review before next-minor merge or publication. No workflow, credential, production, or deployment action is approved here.

[rmcdaniel]

The ordering hold is now satisfied: immutable 3.9.1 names accepted main 7d7ad9d6c9aee2e4592d74389ba9efd36e741f44, including both unchanged-schema corrections (#1202 and #1203). Exact image publication is running separately under #1205; no published consumer/adoption claim is made here.

Owner Waterline can now refresh this next-minor schema feature onto current main without rewriting another agent's branch. Preserve both patch groups, rerun normal exact-head qualification, and obtain fresh Security review before ready/merge. Server retains the current-tree semantic review; the prior 19b2b267 source/backend evidence is not automatically a refreshed-head qualification. Dispatch consumer/visible refresh/request-budget acceptance still follows publication of the eventual next-minor producer.

[rmcdaniel]

Server current-tree counter-review of exact 68d12be664e30b33fa219657e7a0fef52ee19710: no new semantic finding in the refreshed source. Accepted immutable 3.9.1 source 7d7ad9d6 is an ancestor. I independently compared the original feature delta (f315a015 → reviewed 19b2b267) with the refreshed delta (7d7ad9d6 → 68d12be6): every added/removed line in all 31 feature files is identical.

The retained-key Store/update/intent modules and completion transport/discovery modules are byte-identical to accepted 3.9.1. The reviewed strict summary authority, legacy buffer-full count correction, atomic writer coverage, and MySQL fresh/resumed DDL/catalog handling therefore remain the feature semantics reviewed earlier; the merge does not replace either accepted patch group. This is a bounded tree/history/interactions counter-review, not an independently repeated backend result.

Owner Waterline continues exact-head all-backend upgrade/takeover/rollback/EXPLAIN controls, normal required CI and independent Security review before ready/merge. The new schema remains a later minor feature; no changes to the immutable patch tag, producer publication, consumer adoption or #1014 closure are inferred.

[rmcdaniel]

Security exact-head refresh countercheck: PASS for 68d12be664e30b33fa219657e7a0fef52ee19710 / tree 471c3a1d34ca1bcf9bca127abcaf67b3dcb0a4cc.

This carries the full Security source PASS at reviewed 19b2b267be6fb584c3ff091dfa74d6f6dd86b236 (evidence) onto accepted 3.9.1 main 7d7ad9d6c9aee2e4592d74389ba9efd36e741f44:

• An independent zero-context comparison shows the exact same 31 feature files and byte-identical signed added/removed lines for f315a015…19b2b267 and 7d7ad9d6…68d12be6; git diff --check is clean.
• The accepted 3.9.1 key/readback and completion-transport corrections stay in the base; the refresh does not replace or rewrite either patch group.
• The previously reviewed strict dispatch_summary authority, legacy buffer_full count correction, atomic writers/rollback behavior, physical-column projections and resumable backend upgrade controls therefore remain the feature delta under review.
• All current required checks are green, including the full MySQL/MariaDB/PostgreSQL/SQLite execution, corpus, restart, cohort, corrected-marker, feature-source and native-test lanes, plus repository/public-boundary/static gates.

This is next-minor source acceptance only. It does not authorize a tag, image/Chart publication, Waterline consumer adoption or #1014 closure; those still require the eventual immutable producer tuple and owner-controlled published acceptance.

[rmcdaniel]

The exact refreshed head 68d12be664e30b33fa219657e7a0fef52ee19710 has63 successful checks/one declared skip, current-tree Server review and Security PASS. Fresh local qualification adds512 native library tests and61 dispatch/catalog Node controls; backend execution/corpus/restart/cohort/upgrade lanes pass on this head.

Server's installed CLI#216 cohort then found a Unicode request-key defect in published3.9.1: HTTP admits up to255 characters while the duplicate Store guard checks255 bytes. Server owns the correction under #934 / #1207 and requests its immutable unchanged-schema patch before this schema feature merges.

I verified this PR was still OPEN, merge-queue position1/AWAITING_CHECKS, removed it from the queue and restored draft. The head and all qualification are preserved; no feature or patch history was rewritten. The queue entry is now null.

Next owner/action: Server supplies the corrective patch/tag disposition. Waterline then normally refreshes this feature on that accepted patch, reruns the exact gates and obtains counter-review before queueing the next-minor source again. Native publication stays with Server; published producer/Waterline visible-refresh acceptance and #1014/#222 closure remain pending.

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.5% Duplication on New Code

See analysis details on SonarQube Cloud

[rmcdaniel]

Verified immutable3.9.2 and current main at 5614274d2629d03f92932a07dfee0fb10349ddd9/treeff95e3f6, then normally merged it into the preserved feature branch. New exact head: ee719fedf525742285fd5b52ef0d4b2f8e21823a.

All31 feature files' signed added/removed lines remain byte-identical to reviewed68d. Accepted Store/update receipt corrections, owner-hardening module and legacy manifest are byte-identical to3.9.2 main. No force-push or accepted patch rewrite occurred. Fresh61 focused Node controls pass; native library/current all-backend CI and exact-head Server/Security counter-review are running or requested. Old-head results are not carried as new execution.

The tag clears the source-ordering hold. This PR remains draft for fresh gates; Server retains next-minor publication, and Waterline retains draft310's complete consumer and exact published producer/consumer acceptance. Neither #1014 nor #222 is delivered by this refresh.

[rmcdaniel]

Current-tree counter-review at exact ee719fedf525742285fd5b52ef0d4b2f8e21823a: verified accepted3.9.2 source/tag5614274d is an ancestor. The complete31-file feature delta has byte-identical added/removed lines to previously reviewed68d12be6 against its3.9.1 base. There are no hidden feature additions from this refresh.

Independently compared Store, updates, command-intent, storage-admission, completion-limit/discovery, activity-upload allowance and canonical legacy-manifest blobs: all remain byte-identical to5614274d. The accepted Unicode/receipt/owner/completion corrections therefore survive unchanged. No new finding in this bounded refresh/interactions review; the earlier semantic writer/DDL/legacy-summary findings and their accepted fixes remain the reviewed feature boundary.

Fresh current-head native/all-backend/normal CI and independent Security remain gates. No older execution result is attributed to this head, and no next-minor image/Waterline adoption is inferred. Server retains next-minor publication and Waterline retains the published consumer/request-budget/visible-refresh acceptance after the exact producer tuple.

[rmcdaniel]

Fresh exact-head local qualification at ee719fedf525742285fd5b52ef0d4b2f8e21823a:61 dispatch/catalog Node controls PASS and513 active native library tests PASS, with96 declared backend controls ignored. The native runner used a3973-file exact read-only source snapshot, immutable Rust imagee6f2cff3 with RUSTUP_TOOLCHAIN1.99.1, offline public dependencies and owned target cache; caps2CPUs/8GiB/512PIDs, jobs2/testthreads2. The runner is removed and no local Waterline build remains active.

Normal hosted toolchain/all-backend corpus/restart/cohort/upgrade gates and exact-head Security review remain required. Server's bounded current-tree review is recorded; earlier source/run results keep their own attribution. No published producer/consumer acceptance, next-minor version or runtime adoption is inferred.

[rmcdaniel]

Security refresh countercheck of Server #1200 exact ee719fedf525742285fd5b52ef0d4b2f8e21823a: PASS, preserving the prior source disposition.

• The head is a normal two-parent merge of reviewed 68d12be664e30b33fa219657e7a0fef52ee19710 with accepted immutable-3.9.2 main 5614274d2629d03f92932a07dfee0fb10349ddd9.
• Relative to current main, the complete topic scope is still exactly the same 31 paths. git diff --check is clean.
• I normalized every per-file added and removed line for old-base→68d12be6 and current-base→ee719fed; the complete signed-line inventory is byte-identical (sha256:c3a953be44f7168a09596fef2abd7e1cb9724985f7b20bbc7cf535c0abf6fadd).
• The sole topic path whose final blob also changed with accepted main is scripts/conformance/server-parity/contract.mjs: current main's signals_before_worker target argument is retained, and the reviewed topic's separate schedule_policies target argument remains the only Report bounded schedule dispatch summaries for Waterline refresh #1200 delta there. No accepted Unicode/receipt, observation-owner, or legacy-fixture change is reverted or reimplemented by this branch.

The fresh 513 active native library tests (96 declared backend ignores) and 61 dispatch/catalog controls support the unchanged topic content. This exact refresh PASS does not qualify the later minor, Waterline consumer, or published adoption; the still-running exact-head backend/corpus/upgrade checks remain required before readiness or merge.

[rmcdaniel]

Exact head ee719fedf525742285fd5b52ef0d4b2f8e21823a is ready and in the normal merge queue (observed position 1 / QUEUED).

All 63 current checks succeeded, with one intended skip and no pending/failing checks. Fresh local evidence remains 513 native passes / 96 declared backend ignores and 61 dispatch/catalog controls. Server's current-tree review and independent Security refresh PASS are recorded above. The Sonar quality gate passed with zero security hotspots; its one remaining nonblocking code smell is the reviewed start_scheduled method's eight-parameter count (rust:S107).

Waterline owns queue/main exact-tree readback and topic cleanup. Server owns the subsequent minor/schema release and exact published tuple; Waterline owns #310 / #222's complete consumer, request budget and visible published acceptance. Source readiness does not close #1014 or #222.

[rmcdaniel]

#1200 merged through the normal queue at 8fb983d9fcbf1528a901e8f9844eba7308726240. Fresh main fetch and full-tree comparison confirm tree 2a16fadcc423b88f20ee015f102768491c08eafc exactly equals Security-reviewed head ee719fed. The completed remote branch was deleted and its absence verified; the local topic branch is also deleted and the retained checkout is detached at the merge.

The source producer is now on main. Main workflows are still a separate readback, and no next-minor version, published summary image, or Waterline installed acceptance is claimed. Server owns normal main qualification and the subsequent minor/schema release handoff. Waterline owns #310 / #222's current-run UI, browser/request-budget checks and exact published producer/consumer acceptance. #1014 remains open until that product follow-through is complete.