dohertj2/ScadaBridge
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

docs(audit): ScadaBridge C7 review — correct 'six persisted' computed-col wording (5 persisted + IngestedAtUtc non-persisted) + stale perf iteration comment

Browse Source
This commit is contained in:
Joseph Doherty
2026-06-02 15:08:49 -04:00
parent 635461c0fd
commit bc0e5bfd37
4 changed files with 14 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests/Migrations/CollapseAuditLogToCanonicalMigrationTests.cs
+5 -4
View File
@@ -6,7 +6,7 @@ namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests.Migrations;
/// <summary>
/// C7 (Task 2.5) data-projection tests for the <c>CollapseAuditLogToCanonical</c>
/// migration. Verifies that the canonical column layout and six persisted computed
/// migration. Verifies that the canonical column layout and the five persisted computed
/// columns are correct after the migration has been applied:
/// <list type="bullet">
/// <item><c>Action</c> = "{Channel}.{Kind}" per <see cref="AuditFieldBuilders.BuildAction"/>.</item>
@@ -17,9 +17,10 @@ namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests.Migrations;
/// else → <c>Success</c>.</item>
/// <item>Empty <c>Actor</c> maps to NULL.</item>
/// <item><c>DetailsJson</c> produced by <see cref="AuditDetailsCodec.Serialize"/> round-trips
/// correctly and the six persisted computed columns (<c>Kind</c>, <c>Status</c>,
/// correctly and the five persisted computed columns (<c>Kind</c>, <c>Status</c>,
/// <c>SourceSiteId</c>, <c>ExecutionId</c>, <c>ParentExecutionId</c>) resolve to
/// the expected values via <c>JSON_VALUE</c>.</item>
/// the expected values via <c>JSON_VALUE</c> (<c>IngestedAtUtc</c> is computed but
/// non-persisted, so not asserted here).</item>
/// </list>
/// </summary>
/// <remarks>
@@ -30,7 +31,7 @@ namespace ZB.MOM.WW.ScadaBridge.ConfigurationDatabase.Tests.Migrations;
/// seed, apply C5), the tests insert rows directly into the canonical table using
/// <see cref="AuditDetailsCodec"/> — the same codec the migration's
/// <c>FOR JSON PATH, WITHOUT_ARRAY_WRAPPER</c> projection is designed to match —
/// and verify that all six computed columns resolve correctly from <c>DetailsJson</c>.
/// and verify that the five persisted computed columns resolve correctly from <c>DetailsJson</c>.
/// This validates the computed-column SQL expressions that are the source of truth
/// for both the live table and <c>SwitchOutPartitionAsync</c>'s staging table.
/// </para>
tests/ZB.MOM.WW.ScadaBridge.PerformanceTests/AuditLog/HotPathLatencyTests.cs
+2 -2
View File
@@ -29,8 +29,8 @@ namespace ZB.MOM.WW.ScadaBridge.PerformanceTests.AuditLog;
/// <b>C7 re-baseline rationale.</b> The canonical <see cref="ScadaBridgeAuditRedactor"/>
/// works on a single <c>DetailsJson</c> bag (one JSON parse + one re-serialize per
/// call) whereas the former typed-field approach touched many individual
/// fields. Empirical measurements on Apple M-series (2026-06-02, Release build,
/// 5 000 iterations after 500 warm-ups):
/// fields. Empirical calibration measurements on Apple M-series (2026-06-02, Release build;
/// the CI test itself runs MeasureIterations=2 000 after WarmupIterations=200):
/// <list type="bullet">
/// <item>4 KB body (2 body-regex matches + truncation): p50 ≈ 13 µs, p95 ≈ 14 µs.</item>
/// <item>Small DetailsJson, no redactors (JSON parse + rewrite only): p50 ≈ 1.5 µs, p95 ≈ 2 µs.</item>