[releases/28.x][Subscription Billing] Performance: Billing document creation does not scale to large datasets (4M+ subscription lines)

[djukicmilica]

Backport of #8553 to releases/28.x

Backports #8553 — [Bug][SubscriptionBilling]: Performance: Billing document creation does not scale to large datasets (4M+ subscription lines) — to releases/28.x.

Routing analysis (deterministic, per file)

All 14 changed files were classified against real repo state (git cat-file): every file already lives in the BCApps submodule (src/) on the target branch and none exists in NAV App/. Verdict: BCApps only — a same-repo BCApps backport (no src/→App/ remap, no NAV changes required).

How it was applied

• Net PR diff (merge-base ad1c164..PR head) applied onto releases/28.x with git apply --3way in an isolated worktree.
• 13 of 14 files applied cleanly.
• One conflict in GenericConnectorProcessing.Codeunit.al was resolved deterministically by cross-referencing the conflict against the commit diff: the PR's begin / this.ProgressTracker.Finish() / end; wrapper was applied on top of the target branch's existing CreateUsageDataBillingFromTempServiceCommitments call. The compact single-argument overload seen in the incoming side only exists on main (an unrelated refactor not present on releases/28.x), so the target branch's long-parameter signature was preserved. Verified with the conflict-resolver's programmatic checker (PASS).

Backported-from: PR #8553.

Fixes
AB#644742

[github-actions]

Could not find linked issues in the pull request description. Please make sure the pull request description contains a line that contains 'Fixes #' followed by the issue number being fixed. Use that pattern for every issue you want to link.

[alexei-dobriansky]

Agentic PR Review - Round 1

Recommendation: Request Changes

What this PR does

This backport brings the Subscription Billing scale fixes from PR #8553 to releases/28.x. It adds a reusable progress tracker, adds keys, reduces repeated reads, refreshes only affected Recurring Billing groups, changes commit checkpoints, and updates usage-data billing links in hotter paths.

The scenario is valid and important: the linked bug and GitHub issue describe billing runs with about 4 million subscription lines, long runtimes, missing progress feedback, and full rollback after late failures. The code changes mostly match that root cause, and the release-branch conflict in usage-data billing keeps the releases/28.x method signature. The remaining problem is proof: this is a broad billing and usage-data change, but no test files are included.

Suggestions

S1 - Add functional regression tests
Add or backport tests for proposal creation, document creation, usage-based lines, and the new commit behavior. Verify customer and vendor documents, amounts, document links, and usage-data links. This area is financially sensitive, and the existing Subscription Billing test suites make this feasible.

Risk assessment and necessity

Risk: The regression surface is high because the PR changes billing proposal creation, customer and vendor billing document creation, usage-data billing links, page refresh behavior, added keys, and transaction boundaries. Public table procedures on Usage Data Billing are also made externally callable, so the API surface grows in this release branch. I could not find an [AI-REPRO] ADO comment; the detailed scenario comes from the linked GitHub issue and the copied ADO bug.

Necessity: The change is needed because the current flow does not scale to very large subscription-line datasets and can lose hours of work after a late failure. The scope is reasonable for the reported scale bug, but it needs regression tests before merge because incorrect billing amounts or usage links would be costly.

---

[AI-PR-REVIEW] version=1 system=github pr=9847 round=1 by=alexei-dobriansky at=2026-08-03T10:08:13Z lastSha=0617284445c976224f8d81111a97b50a50d3379a suggestions=S1