Skip to content

[Expense tests] Preserve negative fixtures and posted report identity - #11452

Open
Prangshuman Das (t-prda) wants to merge 27 commits into
features/646383-expense-permission-testsfrom
features/646383-expense-negative-fixtures
Open

Prangshuman Das (t-prda) wants to merge 27 commits into
features/646383-expense-permission-testsfrom
features/646383-expense-negative-fixtures

Conversation

@t-prda

@t-prda Prangshuman Das (t-prda) commented Sep 14, 2026 •

Copy link
Copy Markdown
Contributor

Scope

AB#646383

Negative-fixture layer above #11451 in native stack #11893.

Re-enable exactly five CU 148339 methods: four DeleteTravelRequestWith*IsBlocked cases and TravelRequestAPIDatesKeepValidation. The two posted-deletion cases overlap upstream AB#650277.

Upstream now contains the expected-error fixture commits and Last Posting No. lookup. This layer therefore keeps those changes rather than duplicating them.

The remaining posted-deletion fix uses a separate CreatePostedTravelRequestReportForDeletion fixture: a normally posted zero-amount line creates real posted history while preserving the zero-net-spend precondition. Only the two deletion tests call it.

The upstream shared posted-report fixture is unchanged. Its six new lifecycle/reapproval callers and two-line history assertions retain their original behavior and exclusions. This PR does not silently claim ownership of those additional AB#650277 cases.

Validation

Exact parent ancestry, file scope, whitespace and exclusion ownership checks pass. This layer's added/removed source lines are unchanged relative to its refreshed parent. Upstream changes are inherited through the main merge, not reverted. Earlier exact-head cumulative artifact evidence is historical after this parent reconciliation; fresh new-head runtime validation is pending. Existing excluded PDF, IN/RU gaps and tolerated-native distinctions remain; no absent/excluded case is claimed passing.

Validation limits

The previous UserPassword HTTP 401/200/401 result is historical after this reconciliation. CU139496 MicrosoftAuthenticationRespectsServerAuthMode still executes all three requests; Windows200/200/200 runtime remains unverified. The workflow clean-codeunit gate stays default-off and uptake stays explicitly enabled. No provider, authentication contract, new public event, NAV selector or foreign NST change was introduced. Excluded PDF cases and absent/excluded country/native cases remain unverified; prior tolerated-native results are not universal passes. The 59 owned re-enabled methods cover the reviewed fixes, not59 distinct product defects. Validation drafts remain Do Not Merge, outside native stack #11893.

Current checkpoint

Head f3b84be776a95f01d73b3d15c11095e9807b03ba, tree 23be54705318d71569258f9c1f5ca2fecd73f397; parent 460383d69ec0dbf780a9501bb2ba6cf696d2c76c.

Forward-integrated captured main bb7111877ff786951b86a1a0f80d8b39b8f5dacd, including upstream CLEAN27 removal 82b11d26c073de93df3aab17434069f72feab640 (PR12066), to align the direct stacked-PR warning gate. The prior direct uptake checkout retained obsolete source while warning-reference run37020063048 used main73d5794e; RU and CH therefore each reported51 additional warnings (36 AA0244,15 AA0218). Main-targeted RU validations already passed with cleaned source. No warning suppression, parameter rename, partial cherry-pick, or comparator change was made.

Local Pester:117 passed, zero failed/skipped at exact workflow, uptake and full heads. Every layer retains its exact owned patch; all changed baseline blobs equal captured main, and all remaining blobs—including exclusions—are unchanged. The23 committed wrappers/shared success-only finalizer, generator removal, current-process credential identity, ACLs, buffer clearing and narrowed platform classifier remain intact; ordinary configured reruns are unchanged.

Accepted cleanup limitation: failed/cancelled runs rely on normal container teardown; no hard-runner-loss guarantee. Normal CI uses per-run disposable credentials, but supplied credentials may differ. Hook logs prove invocation, not necessarily explicit deletion if teardown already removed the file.

Uptake retains the five query-safe URL compositions in CU148343 StandardSubmissionExposesPolicySnapshot, all assertions,8 methods, setup restoration and the single unlimited-approval fixture. Workflow remains default-off; uptake enables clean execution. Auth visibility/provider contracts and HTTP scenario are preserved. Upstream shared Spend Request zero-amount UnitTest semantics and permission cleanup from PR11561 remain unchanged. CU139806 TestGetCompanyAndEnvironmentDescriptions stays excluded pending its original rationale/current NAV verification (PR11741).

Fresh exact-head GitHub CI is pending, not passed. Previous runs are historical for these new commits. Verify all8 activity methods across22 countries if present; Windows runtime and excluded PDF/native coverage remain unverified. Only this captured main was integrated—no repeated baseline chasing. Merged prefix/native stack #11893 and draft/ready states remain unchanged; validation drafts remain Do Not Merge.

Dedicated follow-up to API auth PR #10085 after current-main run34867798875 confirmed these methods fail. Preserve the tested business/permission contracts; re-enable only the methods repaired by this layer. Static review passed; AL runtime validation remains pending CI.

AB#646383

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Preserve the preceding layer and re-enable only the five fixture methods owned here.

AB#646383

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@AndersLarsenMicrosoft Anders (AndersLarsenMicrosoft) added SCM GitHub request for SCM area and removed Team: Finance GitHub request for Finance area labels Sep 17, 2026
@github-actions github-actions Bot added the Team: Finance GitHub request for Finance area label Sep 17, 2026
…coverage

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@t-prda
Prangshuman Das (t-prda) removed this pull request from stack #11232 September 24, 2026 15:27
@t-prda
Prangshuman Das (t-prda) added this pull request to stack #11863 September 24, 2026 15:27
…rastructure

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@t-prda
Prangshuman Das (t-prda) removed this pull request from stack #11863 September 25, 2026 08:46
@t-prda
Prangshuman Das (t-prda) added this pull request to stack #11893 September 25, 2026 08:47
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@github-actions

Copy link
Copy Markdown
Contributor

Issue #11561 is not valid. Please make sure you link an issue that exists, is open and is approved.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@t-prda
Prangshuman Das (t-prda) marked this pull request as ready for review September 28, 2026 11:15
@t-prda
Prangshuman Das (t-prda) requested a review from a team September 28, 2026 11:15
@t-prda
Prangshuman Das (t-prda) requested a review from a team as a code owner September 28, 2026 11:15
@alexei-dobriansky

Copy link
Copy Markdown
Contributor

Good Sense Reviewer - Round 1

Recommendation: Accept with Suggestions

What this PR does

This change re-enables five Expense Agent tests and gives the two posted-history deletion cases a dedicated zero-amount posted report fixture. The fixture keeps real posted header and line references while leaving the spent amount at zero.

Posting transfers every expense line into posted history before checking whether its amount is nonzero. The deletion tests therefore reach the intended linked-record guards without relying on an invalid negative-line offset.

Problem-solution fit

Fit: Strong

The change directly removes the invalid offsetting-amount assumption and keeps the broader shared fixture unchanged. The other three re-enabled methods already contain the persistence setup needed for their expected-error checks.

Suggestions

S1 (🟠 Moderate): Validate re-enabled tests on current head
Please run the five re-enabled methods on the current PR head. The cited full test run used a different commit and tree, while the current head's PR build did not complete, so the disabled-test removals are not yet backed by exact-head execution.

Risk assessment and necessity

Risk: The code change is limited to test fixtures and the disabled-test manifest, so it does not alter production behavior or public surfaces. The main remaining risk is test reliability because the successful full run used an earlier tree, although the reviewed target procedures remain unchanged.

Necessity: Re-enabling these tests restores coverage for deletion guards and API date validation. The dedicated fixture is narrow and avoids changing production accounting to satisfy a test assumption.


[AI-PR-REVIEW] version=1 promptVersion=4 system=github pr=11452 round=1 by=alexei-dobriansky at=2026-09-28T13:14:21.528Z lastSha=161281001463d24529bc91c3ad0d612229b518a8 reviewKey=d47c1e9388b56d531631740e4ae8d2eeab4dce65e257121e85c2d30ab777b7c5 suggestions=S1@e1be3021

@AndersLarsenMicrosoft Anders (AndersLarsenMicrosoft) added the Team: SCM GitHub request for SCM area label Sep 29, 2026
@github-actions github-actions Bot removed the Team: SCM GitHub request for SCM area label Sep 29, 2026
@AndersLarsenMicrosoft Anders (AndersLarsenMicrosoft) removed the Team: Finance GitHub request for Finance area label Sep 29, 2026
@github-actions github-actions Bot added the Team: Finance GitHub request for Finance area label Sep 29, 2026
…arent

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
…exclusion

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
@alexei-dobriansky

Copy link
Copy Markdown
Contributor

Good Sense Reviewer - Round 2

Recommendation: Accept with Suggestions

What this PR does

The branch was reconciled with its refreshed parent, but the net two-file patch is unchanged from round 1. It still replaces the invalid offsetting-line fixture with a normally posted zero-amount line for the two deletion scenarios and re-enables five tests.

The dedicated fixture continues to create real posted header or line references while keeping the spent amount at zero. This reaches the intended deletion guards without changing production accounting or the broader shared fixture.

Status of previous suggestions
ID Title Status Author response
S1 Validate re-enabled tests on current head Not addressed None. The replacement exact-head build is still in progress, so the five methods do not yet have completed current-head execution evidence.
New observations (commits since round 1)

None - the net PR patch is unchanged and the new commits only reconcile the branch with its refreshed stack and baseline.

Risk assessment and necessity

Risk: Production risk remains low because the net change only updates test fixtures and disabled-test metadata. The remaining risk is test reliability: the five methods are enabled, but their exact-head execution has not completed yet.

Necessity: The dedicated zero-amount fixture is needed to create valid posted history without relying on negative non-correction lines or changing production accounting. Re-enabling the five tests restores useful deletion and API-date coverage once current-head validation completes.


[AI-PR-REVIEW] version=1 promptVersion=4 system=github pr=11452 round=2 by=alexei-dobriansky at=2026-09-30T22:12:33.161Z lastSha=20f1b95d0e533f0eff29c43abb7d4953bc55eeb4 reviewKey=e719ad6b2e550cd315f9e11e7e35c1f5187b3a1cc2584b6c2a78aee5eb75f492 suggestions=S1@e1be3021:notaddressed parentRound=1

Andres Salazar (asalazargeeky) pushed a commit to geekyllc/BCApps that referenced this pull request Oct 1, 2026
microsoft#11654)

## What and why

Consolidate the Expense Agent test repairs on current `main`, following
merged microsoft#11333.

- Fix the three Expense Mgmt. Read/Edit/Admin tests by collecting all
nine permission observations under the exact restricted role, restoring
test permissions, and only then calling Assert. No production
permissions change.
- Replace the invalid ordinary positive/negative posted-history fixture
with a normally posted zero-amount line. This preserves genuine posted
header/line history and zero recorded spend for all eight deletion,
report-recreation, and reapproval cases. Keep identity, ownership,
audit, duplicate-report and zero-spend assertions.
- Re-enable Approve/Submit/Reject Travel Request page-action tests.
Their WebServiceActionResultCode assertion compilation fix is already on
main.
- Classify codeunit 148339 as UnitTest, as requested; it exercises AL
in-process, including direct page procedures. The separate HTTP API
codeunit remains IntegrationTest.
- Repair the foreign-currency fixture in six
[AB#650247](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650247)
API tests (the normal rate, not the adjustment rate, must be non-unit);
refresh cross-session reads and verify persisted identities, dates and
ownership without removing negative operations.
- Remove exactly 14 role/history/action exclusions from the BCApps
Expense Agent disabled-test manifest. No exclusions remain for codeunits
148338 or 148339 in this branch.
- Preserve ALL HTTP API exclusions in BCApps, including the six
[AB#650247](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650247)
cases: BCApps still needs the separate API-authentication prerequisite.
The fixture corrections are retained, but these tests must not be
counted as executed in BCApps. NAV has working API authentication and
enables those six tests separately.

Related:
[AB#650245](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650245),
[AB#650277](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650277),
[AB#625894](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/625894).
NAV test-enablement/uptake also tracks
[AB#650246](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650246)
and
[AB#650247](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650247).

This supersedes the overlapping repair work in drafts microsoft#11451 and microsoft#11452.
Their other already-upstream changes are not replayed. The older drafts
remain open until their owners reconcile/close them after this work is
merged.

## Validation

- Static JSON/exclusion checks and `git diff --check` passed; read-only
review found no significant issue.
- Latest BCApps main was fetched and verified included:
`0c994435e3ee038dbaf66822f8a2791b1aa3f43a`.
- Expected complete BCApps Default suites on this source: 20 tests in
codeunit 148338 and 74 in codeunit 148339, including all 14 re-enabled
methods. HTTP API tests remain excluded in BCApps by explicit user
decision; their runtime evidence must come from the manually queued NAV
buddy build instead.
- Local W1 tenant2-1 resolution initially timed out because MSSQLSERVER
was stopped. After the user requested local repair, that stopped
dependency was started and both relevant databases are ONLINE. Tenant2-1
remains Failed; its Base/System/Application/Expense/Test apps are
uninstalled and Cleaned. Normal tenant synchronization reports the
tenant is not mounted. Shared NST is still 29.0.54137.0, incompatible
with a current full BaseApp30 build.
- No local publication or tests have run. No running service was
restarted, no tenant1-1 data/configuration was changed, and no version
metadata, reset, or reprovisioning was performed. Further local repair
awaits explicit approval for an isolated BC30 environment or a
shared-environment upgrade.
- API authentication uses the existing helper unchanged. PR microsoft#10085
contains a separate shared authentication-provider change; it is not
silently imported here. Until that dependency lands, BCApps API
exclusions remain identical to main. NAV buddy results must establish
whether the historical request-disappearance symptom is resolved.
- CI compile, publication, and actual per-method XML/log execution
verification are pending. A green aggregate check alone is not
sufficient: earlier CI could report success after a codeunit
runtime-compilation failure.
- Private NAV uptake draft: [NAV
#254455](https://dynamicssmb2.visualstudio.com/Dynamics%20SMB/_git/NAV/pullrequest/254455).
It removes 21 NAV method exclusions, including the six API cases, and
restores the Expense Agent Tests app to `Get-AppsToInstall` by removing
the blanket runner filter introduced in
`f09e1491f2adacca7a851a12587966daf91d993d`.
- The user will queue its buddy build manually. NAV remains draft until
BCApps merges, the pointer is updated to an actual main commit, and
buddy logs prove app installation and per-method execution. Static
runner selection is not runtime installation evidence.






Whole-app NAV test exclusion and installation/execution validation:
[AB#650370](https://dynamicssmb2.visualstudio.com/1fcb79e7-ab07-432a-a3c6-6cf5a88ba4a5/_workitems/edit/650370).
The scoped bugs 650245, 650246, 650247, 650277 and 650370 are Active
while verification is pending.


## CI iteration: 79ef58 (superseded by RU fixture correction)
Actual W1 Default JUnit artifact 10646077590 from run35596801127 records
148338 **20/20 passed** and 148339 **74/74 passed**, zero errors/skips,
including all14 re-enabled methods. DE Default JUnit also
records20/20+74/74.

RU Default job106353051317 failed: 14833820/20passed, but all74 Spend
Request tests failed during Initialize because
PrepareNormalGenPostingSetup could not find a fully populated General
Posting Setup in the UnitTest database. This is fixture initialization,
not74 different product failures.

Commit0ed0d44f92 creates a complete posting setup using existing Library
- ERM helpers before the country-specific setup. New-head validation is
pending; the prior-head W1/DE passes are not a claim that this new
commit has passed. No tests were disabled for this failure. Local and
NAV buddy execution caveats above remain.


## CI iteration: 0ed0d44 (superseded by CH VAT fixture correction)
Actual JUnit from run35616107333 attempt2 verifies **W1 and RU each
passed all20 permission tests and74 Spend Request tests**, zero
failures/errors/skips. W1 artifact10658609682; RU artifact10658544297.
This confirms the RU posting-setup correction.

CH Default job106447478820 failed: permissions20/20passed; all74 Spend
Request cases stopped in country VAT initialization because the selected
Normal VAT template referenced a nonexistent Sales VAT Account.
Artifact10659053532 and the job stack identify Library - ERM Country
Data.CreateVATPostingSetup.

Commit bd80ea3 initializes the VAT template before country setup,
creating replacement Sales/Purchase VAT accounts only when the
referenced account is missing. Valid account references and all
assertions remain unchanged; no test is disabled. Fresh validation of
W1, RU and CH on this new head is pending. Older-head pass evidence is
not a claim that the new head passed.


## Prior-head test execution: bd80ea3
Downloaded actual JUnit artifacts from run35644720549 attempt2 and
verified every required re-enabled method, not just the job conclusion:

| Country | CU148338 Expense Permissions | CU148339 Spend Request |
Artifact |
| --- | --- | --- | --- |
| W1 | 20/20 passed | 74/74 passed | 10668715386 |
| RU | 20/20 passed | 74/74 passed | 10667802737 |
| CH | 20/20 passed | 74/74 passed | 10668516148 |

All three have zero failures, errors and skipped cases in these suites.
All14 re-enabled role/posted-history/page-action cases executed. This
verifies both country fixture corrections and placement in the
Default/UnitTest lane. It does not establish fully test-owned VAT setup:
the current helper repairs the selected existing VAT template when
necessary.

Run35644720549 attempt2 completed with160 successful jobs,2 skipped and2
failed (BE Uncategorized plus the aggregate status check). The
status-check log names ONLY the BE platform/indexes/platform.json
download/setup failure; no Expense test failure remains in the verified
W1/RU/CH suites. The overall CI gate is therefore still red. No rerun
was initiated by this agent. HTTP API tests remain intentionally
excluded in BCApps; current NAV buddy and local runtime validation are
not claimed. The user-supplied older NAV buddy3641608 did install
Expense Agent and its test app and pass20 permission tests, but all74
Spend Request cases hit the pre-fix RU initialization failure; its queue
timestamp preceded the RU correction.


## Current fixture repair: 2d45e6f
The supplied NAV buddy3641753 AU log passed all20 permission tests but
failed all74 Spend Request tests in
UpdateZeroVATPercentInVATPostingSetup: the demo-data helper assumed an
existing blank-business-group VAT row. Earlier BCApps AU also
passed94/94 on bd80ea3; that baseline difference does not prove the
fixtures are independent.

Replace broad demo-data normalization rather than adding another
country-specific fallback:
- Create fresh Expense Posting Groups and their G/L accounts, fresh
Employee Posting Groups with valid expense accounts, and fresh payment
methods. Bind the report fixtures to these records instead of reusing
arbitrary posting masters.
- Remove the unrelated country VAT/general/purchase setup mutations and
the earlier RU/CH template repairs. These Spend Request lifecycle
fixtures use no-VAT expense accounts; retain the explicit country helper
for Journal Templ. Name Mandatory=false.
- Restore General Ledger, Human Resources, Source Code and Expense Agent
singleton setup between tests with Library - Setup Storage. Retain
existing scoped Expense cleanup; do not delete unrelated VAT or ledger
masters.
- Preserve all74 test methods, assertions, scenario/feature tags and
handlers. No exclusions or production behavior changed.

Validation: static call-site checks and git diff --check passed. Fresh
local resolution now reports tenant2-1 Operational, but a dedicated
tenant2-1 AL MCP download returned404 for System30.0.0.0 (No published
package matches the provided arguments). Application symbols downloaded
to the worktree/tenant cache; required symbols remain incomplete. No
local compile/publication/test was performed, and no shared symbols,
metadata changes, reset, service restart or tenant1-1 modification was
used. Fresh CI and manually queued NAV buddy execution are pending; all
bd80ea3 pass evidence above is prior-head only.

Pipeline audit: both NAV and BCApps AU UnitTest setup use an empty
company, not Contoso generation. BCApps can retain the Disabled runner
across ordinary app passes; the exact origin of the differing AU VAT
data is unproven. Open PR microsoft#10085 has overlapping
clean-tenant/split-isolation remediation. No shared-runner changes are
imported into this fixture PR.


## Warning-gate correction: 67d0e3c
Run35714820873 on2d45e6f finished failed. RU Default and SE/IT Clean
logs identify two new AA0198 warnings: the new global
LibraryHumanResource shadowed two existing local declarations in
SpendRequestTest. RU stopped at the compilation/new-warning gate before
publication and tests; this was introduced by the fixture change, not
infrastructure.

Commit67d0e3c removes exactly those two redundant local declarations and
reuses the global instance. All74 test bodies/assertions are unchanged;
exact two-line diff and diff checks passed. Fresh CI compile/publication
and actual W1/RU/CH/AU94-test execution remain pending. Monitoring
resumes every30minutes; no buddy build or CI rerun is automatically
queued, and local tenant-specific System30 symbols remain unavailable.


## Payment-method fixture correction: 87357b7
The67d0e3c app builds passed for W1/RU/CH/AU, clearing AA0198. Actual AT
Default JUnit artifact10705006253 from run35730199203 records
CU14833820/20passed and CU14833955/74passed,19failed. All19 failures are
the same fixture defect: Expense Payment Method enforces uniqueness by
nonblank Reimbursement Type, so creating a new Employee Paid method for
each report collides with the prior method. This is not a product
assertion failure.

Commit87357b7 clears only Employee Paid/Company Paid payment-method
fixtures in Initialize, after Expense transaction cleanup and before the
initialized guard. The six factories use the existing find-or-create
library helper: the first call creates this test's method, and further
calls reuse it without violating uniqueness. Posting-account ownership
and all74 scenarios/assertions/tags/handlers remain unchanged. Static
checks verify every-test cleanup ordering, all six callers, preserved
assertions and no HR-library shadowing; runtime verification on this new
head is pending.

The earlier run also has an independent LegacyTestsBucket1
container-setup failure downloading platform/indexes/platform.json
(HTTPS connection timeout). No automatic retry or
shared-runner/environment change is included. Local tenant-specific
System30 symbols remain unavailable; latest NAV buddy execution remains
unverified.


## Verified current-head runtime: 87357b7
Downloaded actual JUnit from run35751572706 attempt1 and checked
complete suite counts, zero failures/errors/skips and every required
re-enabled method:

| Country | CU148338 Permissions | CU148339 Spend Request | Artifact |
| --- | --- | --- | --- |
| W1 | 20/20 passed | 74/74 passed | 10714234115 |
| RU | 20/20 passed | 74/74 passed | 10714194977 |
| CH | 20/20 passed | 74/74 passed | 10715030599 |
| AU | 20/20 passed | 74/74 passed | 10714392965 |
| AT | 20/20 passed | 74/74 passed | 10713924969 |

All14 re-enabled methods executed in each country. This verifies the
latest payment-method cleanup/reuse correction, including the AT
regression. All five country app builds passed. W1 job106867710487
additionally records publication, synchronization and installation of
both Expense Agent (Preview) and Expense Agent
Tests30.0.2147483647.85851, followed by successful execution of both
codeunits.

The overall workflow is not green: it is still running, with US
LegacyTestsBucket2 job106867713372 failed because its self-hosted runner
lost communication with GitHub. That infrastructure failure is separate
from the verified target suites; no automatic rerun was requested.

NAV draft254455 remote commit e939957a13d896b18ae1bf053a14ab6949a052fd
already consumes this exact87357b7 commit. Fresh NAV buddy verification
remains necessary, especially Codeunit isolation and the six NAV-enabled
HTTP API tests that remain excluded in BCApps. The supplied older
NAV3642277 W1 log references67d0e3c and reproduces the now-corrected19
payment-method failures; it does not test87357b7. No current NAV/local
runtime pass is claimed.


## Final BCApps CI gate
Run35751572706 subsequently completed SUCCESS on attempt2 for the
same87357b7 head:162 successful jobs,2 skipped,0 failed. The earlier US
legacy-runner communication failure is cleared. The agent did not
request this retry. Five-country actual20+74 execution evidence above
remains valid for this exact source commit. NAV validation remains
separate.


## NAV HTTP API follow-up: af10e62 (validation pending)
The supplied NAV3642786 NL IntegrationTests log (build30.0.54998.2786)
executes all15 CU148347 cases:7pass,8fail. These HTTP API cases remain
intentionally excluded in BCApps, so the prior87357b7 unit-test/green-CI
evidence does not validate them.

Library - Graph Mgt. appends successful responses to the supplied
ResponseText and raises failed requests before populating that output.
The failures expose two test-harness defects: stale concatenated
responses explain the missing traveler expansion and two
foreign-currency assertions; parsing empty response JSON explains four
negative-case failures. This test-only commit clears all32 HTTP response
buffers and checks failed requests through the helper's raised error,
preserving expected HTTP400/ALDialog, domain/owner/request-number
details, the server BadRequest read-only-status guard, and persistence
checks. All15 methods, existing tags and conditional compilation remain
unchanged.

The eighth failure is still under investigation: the dates case fails
the first local GetBySystemId immediately after successful POST, before
PATCH. The test now explicitly compares the returned id with the
client-supplied id before the existing persistence check, so a future
failure can distinguish rewritten/ignored identity from record
disappearance. The supplied-GUID requirement is NOT removed and no
production workaround is applied.

Static diff/buffer/case checks pass. A fresh dedicated tenant2-1 symbol
download retrieved5of6 requested dependencies but again failed404 for
System30.0.0.0; no local compile, publish or tests ran. Fresh CI
compilation is pending and cannot substitute for NAV API execution while
BCApps exclusions remain. Another NAV buddy build has not been queued
automatically; the prior queue authorization was used for2773217. Both
drafts remain in-progress, not fully validated.



Current follow-up head: ff267e3. The
returned-identity diagnostic uses the existing AL Evaluate(Guid,
JsonValue.AsText()) pattern. This supersedes af10e62; its identity
investigation and runtime-validation caveats still apply. No additional
NAV build has been queued.




## ff267 CI gate completed
Run35869253473 attempt1 completed SUCCESS on
ff267e3:162 successful jobs,2 skipped,0
failed. Actual W1 and RU build logs confirm Expense Agent Tests
compiled, with unchanged warning baselines (40304 W1;42672 RU). CU148347
HTTP API cases remain excluded in BCApps: this verifies compilation and
the CI gate, NOT the eight NAV API failures or the unresolved date POST
identity. Local runtime validation remains pending; no NAV rerun was
initiated by this monitor.



## Locally verified API persistence fix: 29d7bb5
The remaining two API failures were reproduced outside the test runner:
user-scoped owner/date-only POST returned201 but no stored request
(blank number; subsequent GET returned0). Adding purpose caused
insertion, but the tests were NOT weakened with such a payload
workaround.

Fix page7134 instead: copy supplied date values onto Rec to register a
real record change while retaining deferred paired validation; clear the
route-prefilled owner on new records and default it at insertion only
when requestedBy was omitted. Explicit supplied owners still undergo
scope validation. Requested identity and all existing
date/ownership/persistence assertions are retained.

Local validation after the explicitly authorized BC30 W1 reset (agents
enabled, two tenants), exclusively on tenant2-1:
- Downloaded fresh tenant-specific symbols; built and published both
Expense Agent (Preview) and Expense Agent Tests30.0.0.0 through a
dedicated AL MCP host.
- CU148347: all15 HTTP API tests PASSED,0failures/0skips, runner130451
(required isolation Disabled), CRONUS W1, password-only test
administrator. Includes complete date-only PATCH sequence and both
original missing-record cases. Evidence:api-local-insert-fix.xml.
- Empty Company regression: CU14833820/20 andCU14833974/74
PASSED,0failures/0skips, runner130450 Codeunit isolation.
Evidence:local-unit-regression-20260924.xml. The helper's 'Disabled'
setup log describes code-coverage tracking, not isolation.
- Additional direct HTTP checks passed: supplied-id/date-only POST
persists; omitted owner defaults from route and persists; explicit blank
owner and unscoped missing owner return400 without inserting.

Local runtime builds use CodeCop/UICop with default severities; partner
AppSourceCop defaults incorrectly reject first-party namespaces/IDs and
the full repository ruleset promotes pre-existing obsolete/style
warnings. No source warnings were suppressed or manifests altered; CI
remains the full first-party compilation gate. Fresh current-head CI and
NAV buddy results are still pending.

BCApps HTTP API exclusions remain unchanged pending the separate
authentication prerequisite; NAV keeps these APIs enabled. The
Windows-linked local account was unsuitable for the existing Basic-auth
helper, so a password-only test account was used on tenant2-1. No
production authentication behavior was changed. Canonical reset-tool
compatibility fixes remain local and are not included in this PR.


## Final merged-source local verification: aaa5a76
Merged current BCApps main (a18b1b3) normally to resolve the sole
namespace-import conflict, retaining both imports and all upstream
changes. Main contributes six additional Spend Request tests, bringing
CU148339 to80; no scenarios were removed.

Rebuilt and published Expense Agent and its tests from the merged
source. The seed Base Application lacked the newly inherited
Purpose-validation behavior, so the matching unmodified Base
Application30.0.0.0 was built from the same source and published only to
tenant2-1 (verified tenant scope). No BaseApp source workaround or
version/metadata override was used.

Final persisted local results on tenant2-1: **15/15 HTTP API +20/20
Expense Permissions +80/80 Spend Request =115/115 passed**, zero
failures/skips. HTTP runner130451 in CRONUS W1; unit runner130450 in
Empty Company. Reports:api-local-final-main.xml
andlocal-unit-final-main.xml. The original date/owner POST payloads and
persistence assertions pass without adding purpose. Source worktree is
clean; compiler-generated report-layout edits were not committed.

This supersedes the earlier unresolved-identity caveat. New-head CI and
NAV buddy results remain pending; BCApps API exclusions and NAV API
enablement policy are unchanged.


## W1 Bucket5 fixture follow-up: 40a64b5
The supplied NAV3644685 log reports4,814passed/2failed:
CU139194.CDSConnectionWizardCheckModifyCDSConnectionURL rejects a
positive fixture outside dynamics.com; CU139148.NotExpiredToken fails
its expiry assertion. These are separate from the repaired Expense APIs.

Test-only corrections: use valid dynamics.com hosts for positive HTTPS
normalization cases, preserve explicit-port/path expectations and HTTP
rejection, and exercise SaaS validation with environment restoration.
JWT fixtures now use the public Unix Timestamp helper rather than manual
timezone/epoch arithmetic, with a one-hour past/future margin instead of
one second. Production URL/security and JWT-expiry implementations are
unchanged; no further tests are disabled.

Both test apps compile locally through AL MCP with tenant2-1 symbols.
The corrected CDS case passes locally. Full local validation is
incomplete: the CRM suite has pre-existing local
AutoRollback/EmailLogging and on-premises-mode setup failures; the
latter reproduces independently of the modified case. Tests-Misc server
publication rejects its existing MockAzureKeyVaultSecretProvider
reference despite successful local compilation. The missing standard
MockTest add-in was restored, but no service restart or validation
bypass was performed. Token runtime and full-suite verification must
come from the buddy build or follow-up local runtime maintenance. No
full-pass claim is made for these two codeunits.

The prior115 local Expense API/permission/Spend passes remain evidence
for unchanged Expense code, not proof of these new fixture repairs.
Existing NAV retention/VIES quarantines and differing BCApps/NAV API
policies remain unchanged.

## Assisted-setup fixture follow-up: c52631a
After ADO2787333 failed, the matching WorkIQ email for NAV3644810
identified two RU Bucket5 failures in CU139196:
RunAssistedSetupFromNormalSetupRecordMissing and
RunAssistedSetupFromNormalSetupRecordExists. Both used TEST as a
positive Server Address and hit the dynamics.com host guard.

Only those two fixture URLs become https://test.dynamics.com, with
corresponding comments updated. The existing wizard-propagation
assertions, handlers, test list and production validation remain
unchanged; no additional exclusions were added.

The CRM test app compiled and published on tenant2-1. Local runtime
verification is blocked before these assertions: unchanged Initialize
calls DisableEncryption, which is unsupported in this agent-enabled
environment (47 reported,4 passed,43 failed in setup). No
encryption/credential/service changes or assertion bypass were
performed. NAV buddy execution is the remaining validation route; these
two cases are not yet claimed passing.



## Scope correction: 5de4cb7
At the user's explicit request, this BCApps PR is limited to Expense
Agent and Spend Request changes. Our earlier changes to
CDSConnectionWizardTests, CDSConnectionSetupTest and UTREST have been
restored exactly to their pre-repair aaa5 contents. The final PR diff
contains only four Expense Agent AL files and its disabled-test
manifest; the historical non-Expense fixture proposals above are
superseded and NOT included.

NAV handles unrelated failures through exact-method temporary exclusions
instead: the four confirmed Bucket5 methods (one CDS wizard URL case,
two CDS assisted-setup cases and UT REST.NotExpiredToken), alongside the
previously authorized four Retention Policy and ten VIES methods. These
are quarantines, not fixes or passing tests; bug filing remains
deferred. All other existing exclusions are preserved.

BCApps keeps the Expense non-API re-enables and its HTTP API exclusions.
NAV retains the Expense app re-enable and its21 method re-enables,
including the six HTTP API cases. The prior
local15API+20permissions+80Spend passes apply to unchanged Expense code;
fresh CI/buddy results remain pending. Future repairs in this BCApps PR
must not modify unrelated tests.

---------

Copilot-Session: 988fb581-008c-4e0d-99c1-285cd45d49e7
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 3952f078-a881-4da8-ad96-13b727e48a91

This branch was successfully deployed

1 active (outdated) deployment
triage — 16128100 Deployed Sep 29, 2026 by AndersLarsenMicrosoft via Classify team ownership #6202
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

AL: Apps (W1) Add-on apps for W1 SCM GitHub request for SCM area Team: Finance GitHub request for Finance area

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants