Repository navigation
[Task] Prepare and diagnose the full consolidated source build #8287
Description
Activity
Build preparation is active in an isolated disposable tree. Fresh upstream main SHAs still match the inventory (Extensions 33fa0bf, Studio 9afd3e3); 17 Extensions and 14 Studio open PR heads were recorded without changing their status. The fresh history rehearsal retains 9,635 exact files and all original ancestors at Core 8e893e0. After scoped build/reference integration, the mapped WorkflowContexts Studio module builds on net10 with zero errors and 32 retained warnings. A full consolidated solution build is running; it is not yet accepted. This also exposed existing Extensions connection-persistence identities that the proposed #8283 implementation must preserve through additive naming.
Preparation and the two verified UI/sample compatibility fixes are proposed in #8291. Full declared-framework build failed with 25 errors; Dapper and MongoDB need real tenant-scoped TryUpdateLatestAsync implementations. Agents now builds on net8/net9/net10, six stale-response validation regressions pass, and the net8 sample builds. The reviewed preparation was replayed in a second clean disposable checkout and those tests/sample build passed again. No synthetic rehearsal commit was pushed; no migration or publishing completion is claimed.
- addedelsa 3This issue is specific to Elsa 3This issue is specific to Elsa 3enhancementNew feature or requestNew feature or requestprio highIs on the roadmap for the near-futureIs on the roadmap for the near-future
on Sep 23, 2026 Build preparation PR #8291 merged as 554b48b after all exact-head checks, 5/5 review and independent review/guard regression (8/8 normal and optimized Python). This delivers the reproducible source/build patch, async Agents validation repair and Blazor sample framework repair. #8287 remains open: Dapper #8293 and Mongo #8294 integration plus the Core caller fix #8292 must be incorporated, followed by a complete declared-framework solution build and final-layout tests. No synthetic rehearsal history was pushed and no package was published.
Build follow-up in PR #8306: the complete mapped rehearsal now builds (0 errors, 1,561 disclosed warnings). The five historical restore-blocked projects execute locally with 107 passing tests and one existing Azure Service Bus skip. Exact revisions, commands and hashed logs are retained. This is an open evidence PR; fresh CI, final imported layout and complete dependency closure remain unaccepted. No package publication.
The consolidated build rehearsal previously used Core before the merged credential lifecycle and EF/BPMN corrections. This adds an exact, guarded Core
076f022profile while retaining the original source pins and recipe, then records its complete 340-project build and targeted integration results. Arbitrary source revisions remain rejected.The build passed with zero errors and 1,867 warnings after an initial disk-exhaustion failure; both attempts and their hashes are retained. Freshly built net10 outputs passed 214 tests across Connections, BPMN and imported Studio Agents. Existing dependency advisories and MySQL/EF version constraints remain explicit limitations. This is a disposable rehearsal, not the actual history import or release approval.
Validation: preparation guards 9/9 normally and with Python optimization; log hashes verified; 90 issue states/titles refreshed and all 89 native parent edges/39 blocking edges verified live through GitHub REST. The hierarchy includes the bounded PostgreSQL provider and separate-worker recovery follow-ups.
Refs #8194, #8287, #8286. These issues remain open pending their remaining integration gates. No package publication, deployment, production change or repository archival.
The combined Core/Extensions/Studio test run found 21 failures because three Studio test projects still locate sources through standalone repository paths. This adds a shared source-file resolver to their integration patch while preserving every CSS, host configuration and module-boundary assertion.
Validation: the initial full run is retained as failed (94 TRX reports: 6,169 passed, 21 failed, 148 skipped). After the fix, all 103 tests in the three affected projects passed, with no skips. A separate clean patch replay reproduced all seven resulting file hashes. The full run was not repeated after this test-only correction; logs and that limitation are explicit.
Refs #8287, #8286, #8194. Apply after the current-Core rehearsal artifacts recorded in #8312. This is a reviewed integration artifact for the eventual source import, not a completed import or publication.
Canonical preparation #8324 merged at
77f3ca92eb3e514af959404f67ea538b07cba98fafter exact-head checks/review. The fresh selected-Core95a rehearsal preserves 6,222 Core and 3,771 mapped upstream files (9,993 exact blobs/modes), with all original histories reachable. Root independently verified those facts.The new packaging-property matrix passed 2,586 cases across 167 imported projects; root independently verified every case, complete Debug/Release × reference-mode × framework coverage, and disabled package generation. The canonical solution contains 343 projects and 95 project names selected by NUKE; 15 Studio test projects live under src/studio, so the receipt's 16-of-17 test-tree subset is not the entire test inventory.
The current 17 Extensions and 14 Studio open PR heads still match the retained upstream-work ledger. This latest read occurred after raw candidate creation and before the canonical build. Their work remains untouched.
A full canonical Compile/Test run is next. An initial command probe was aborted before tests and is retained as an aborted result, not successful validation. The original 076 full-build/test evidence and new95 receipts remain separate. No source import or publication has occurred.
Canonical-source evidence #8328 merged at
6f493809eae0e1652ca185b901a982984f4ef799after exact-head checks, Greptile 5/5 and independent root verification. The selected Core95a / Extensions33fa / Studio9afd rehearsal preserves 9,993 raw blobs/modes and original ancestry. Its canonical 343-projectElsa.slncompleted NUKE Restore/Compile/Test with exit 0: 6,222 passed across executed target frameworks, 148 skipped, zero failed. Fixed TRX naming retained only 6,056 passes because two Designer framework results were overwritten; all three 83-pass runs remain in the hashed log. The 95 selected commands include one benchmark-only project with no tests and two explicitly skipped-only projects. The 21 earlier Studio path failures now pass, plus 32 added tests.The 167-project / 2,586-case property matrix is pre-supplemental-patch evidence; the full build/test receipt includes the four separately hashed compatibility patches. The 1,570 compile warnings and dependency advisories remain recorded. This is a selected-pin disposable rehearsal, not the actual source-history import or a published release. #8287 and #8313 remain open: #8325 activity migration, runtime/Studio proof, asset dispositions, refreshed graph and final imported-head verification are outstanding. No package publication or deployment occurred.
Source-grounded frontend build proof is open in #8334. On an isolated Core 95a658b / Extensions 33fa0bf / Studio 9afd3e3 mapped clone, Node 22.22.1 passed the BPMN generated-type check, 253 Designer tests and type checking, and both Designer/DomInterop webpack builds; six browser assets matched the bounded Workbench/Studio runtime receipt. The PR carries the repeatable command and import-time generator layout patch. #8287 remains open: the history-bearing import and a required CI gate on the imported source have not landed, and the recorded dependency advisories need release review.
Studio ClientLib build integration PR #8334 merged as
a1d102f2e4db929bb554b7a0a715d8d42f8610d2after its exact-head PR checks, Greptile 5/5, and resolved threads. The Linux check built the pinned standalone Studio Designer and DomInterop ClientLibs with reviewed Node 22 lockfiles, generated BPMN types, Designer tests, and all six required browser assets. The same workflow is path-triggered for the mappedsrc/studiolayout after import. The merge title used[skip ci]; no Packages run is present for this merge head and no feed publication occurred.This is a frontend build lane, not acceptance of #8287. The current-tip history-bearing import, full source build/framework matrix, Secrets package/schema compatibility, and imported Workbench browser rerun remain open. The standalone Studio pin differs from the latest upstream main and must be refreshed for the final import.
The first full prepared-source compile is now complete. A disposable, no-remote three-parent rehearsal at Core
a13ac7a412e037280d7a568fd9dd87b06ff8b724, Extensionsba8b71d91c15ffe5be4b2c539cf9f712e74af775, and Studio20ceaeeed7e671f0c9662003e82063026f2216dewas prepared with the reviewed Workbench/Studio fixture overlays. SDK 10.0.300 built the 343-project mappedElsa.slnusingdotnet build Elsa.sln -p:UseProjectReferences=true -p:IsPackable=false -p:GeneratePackageOnBuild=false -m:1: exit 0, 0 errors, 1,824 warnings, 32m32s. The private raw log has SHA-256babcf33dab00ff80be08ae9396c17dcc71bc81e05c03833c75babd5f0bf4ea0e; a sanitized receipt is being prepared for a focused PR.This proves compilation of the prepared mapped source, not a clean imported checkout or release package provenance. The no-remote rehearsal produced SourceLink warnings, and some dependency-audit warnings remain. Targeted runtime/Studio tests, clean package-consumer checks, final history import and publisher cutover remain open. #8287 stays open pending those checks and the reviewable receipt. No package was packed or published.
The sanitized mapped-build receipt and targeted Secrets test evidence are in PR #8380 (head
54303296e75afa0ae4bab1eb2bcbbf8357cac2eb). The prepared 343-project source build passed with 0 errors; mappedElsa.Secrets.UnitTeststhen passed 141/141 on net10.0 with no skips. PR checks and review are still running, so this is proposed evidence rather than merged delivery. #8287 remains open for canonical history-bearing import, full combined tests/framework matrix, clean package-consumer and SourceLink proof, and publisher transition. No package was packed or published.Evidence update (2026-09-24): #8380 merged after exact-head checks and a resolved Greptile 5/5 review. It records the nonpublishing 343-project mapped Core/Extensions/Studio source build at the current source pins, plus the mapped 141/141 Core Secrets unit run. The raw build and test logs remain private; the committed receipt contains hashes and nonsecret summaries. The merge did not trigger Packages or wiki publishing.
This is a prepared, overlayed source-reference rehearsal. It is not the canonical history-preserving import, package graph build, Studio browser proof, or release. Keep #8287 open for those acceptance criteria.
The refreshed source-tip/open-PR disposition receipt is in #8384 (open; review and checks pending). At the 2026-09-24 read, Extensions and Studio main still matched the selected ba8b/20cea import pins; Extensions #216 had merged as the exact selected Extensions tip, and the other 16 Extensions plus 14 Studio PR heads remained open and unchanged. This preserves upstream work for final owner review and does not complete the canonical import or retroactively place the check before the existing synthetic rehearsal.
Evidence refresh merged in #8384 (725f1a6). It records the exact open PR collection at the selected Extensions/Studio pins, including REST page coverage, response hashes, and native-field snapshots. The earlier #8380 mapped 343-project build receipt also remains accepted. This is a point-in-time input to the canonical history-preserving import, not completion of #8287: repeat the upstream collection at import time, resolve overlapping open PRs and source history, then rerun the consolidated build and compatibility checks. No packages were published.
Upstream intake refresh (2026-09-25 UTC) before the history-bearing import:
- Core
main:d2c6830a9578e13ca1ce244eb87c19434f09de29; Extensionsmain:ba8b71d91c15ffe5be4b2c539cf9f712e74af775; Studiomain:20ceaeeed7e671f0c9662003e82063026f2216de. Extensions and Studio still match the selected source pins. Core has advanced beyond thea13ac7arehearsal pin. - Compared with the accepted upstream PR refresh, three new open main-targeted PRs are visible: Extensions #222 at
6af660846ff75c065be23f2e3b681a62f0915a5b(Dapper/Mongo compare-and-swap test coverage), Extensions #223 at9039db3908619945e606d6fc7e5ae784ad5bc841(release-branch PR CI trigger), and Studio #1058 ateb1b8423f5911ad1c9fb41b18dcce12c530483b0(release-branch PR CI trigger). The first changes two test files; the others change CI source/workflow files.
These PRs are not in the selected source tips and have not been merged, rejected, reparented, or imported by this refresh. Recheck their disposition and any newer upstream work immediately before constructing or merging the import. The current full mapped-source evidence remains a preparation/rehearsal, not an accepted history-bearing import or release proof.
- Core
Fresh current-source rehearsal is proposed in PR #8394 (head 8671baa; exact-head CI/review pending). Source pins: Core c4b3ce1, Extensions ba8b71d, Studio 20ceaee. The disposable no-remote synthetic commit cfa5e7d6d6ccc5f849163b3f6e0db651b6062b38 preserves all three histories and exactly maps 10,191 files. The 167 imported/test projects passed 2,586 packing-disabled property evaluations. Clean mapped build failed on four known Workbench Secrets calls; after five reviewed fixture overlays, the 343-project source-reference build passed with zero errors and 1,556 warnings. Focused net10 tests passed: Management 127, Core Secrets 141, Studio Designer 83, Studio Secrets menu 11, Studio Agents 6. PR artifacts contain the source PR refresh, hashes and exact limitations. This is local rehearsal evidence, not the history-bearing import or final release/consumer proof; #8287 stays open.
The bounded current-source build preparation is accepted in PR #8394, merged as
337a2573635f7892ba298115307dad4c9f76b382from exact headbb24b1a71a8693e884c2415155542b4880b6ab51. Exact-head CI passed, Greptile reported 5/5, and the review thread was resolved. The merge used[skip ci]; no Packages or wiki run appeared for that SHA.The retained receipts pin Core
c4b3ce150160e3c9062b57f7b158fd6b968e1631, Extensionsba8b71d91c15ffe5be4b2c539cf9f712e74af775, and Studio20ceaeeed7e671f0c9662003e82063026f2216de. A no-remote disposable history rehearsal mapped all 10,191 source files by exact blob and mode and passed 2,586 packing-disabled property evaluations. The mapped 343-project solution built with zero errors after applying the five separately reviewed fixture patches; focused Core and Studio tests passed. The PR also records the fresh open upstream PR inventory and the unsupported/unknown areas.This closes the preparation and diagnosis task. It does not merge the history-bearing source import, establish final package-only/remote SourceLink proof, complete the Secrets upgrade and browser gates, or authorize publication/cutover. Those remain tracked by #8286, #8260, #8275 and #8326. Recheck upstream tips and rerun the build/proofs at the actual import candidate.
Program: #8194
Parent: #8286
Deliverable
Prepare the explicit import map in an isolated disposable workspace at recorded Core/Extensions/Studio SHAs, preserving all original commit ancestry. Make the mapped source buildable through scoped project-reference, central-build and asset integration changes. Keep the byte-preserving relocation receipt distinct from subsequent integration edits.
Acceptance
Scope boundary
This task establishes a reproducible build candidate and diagnoses compatibility. It does not claim completed migration, approve cutover, invent pilot demand or authorize public package release.
Dependencies
Accepted inventory #8251/#8252 and merged history rehearsal #8268. #8275 and #8260 remain integration gates rather than reasons to defer independent build preparation.