/ 
Test-FitNesse-Slim-1.00
/ docs/fitnesse-acceptance-coverage.md

FitNesse acceptance coverage comparison

Generated from the two committed baselines and the audited FitNesse v20251025 selection. The isolated and native modes remain separate gates.

| Inventory | Pages | |---|---:| | Individually executed | 50 | | Native suite selected | 93 | | Shared | 48 | | Individual only | 2 | | Native only | 45 | | Union | 95 |

Why the inventories differ

The individual runner is a deliberate 50-page regression inventory. It clears generated MySuite pages between requests and checks each selected page in isolation. The native runner delegates traversal to FitNesse's suite responder, which includes enabled empty Test properties, legacy test headers, and referenced support pages. It also preserves suite execution context and ordering.

No filter tags were supplied; Suites properties did not constrain this suite selection. Seven property-marked pages are excluded by their own or an ancestor's Prune property. Selection source counts: Test=true property: 48, cross-reference/support page: 3, empty Test property: 32, legacy Test header: 10.

Independent regression evidence

| Dimension | Isolated pages | Native suite | |---|---|---| | Page lifecycle | One HTTP test per selected page; generated MySuite is cleared between requests. | FitNesse traverses and executes the complete selected suite in order, retaining shared suite context. | | Discovery | Deliberate 50-page regression list with explicit Test properties. | FitNesse v20251025 suite traversal, including empty Test properties, legacy headers, references, and pruning. | | Primary evidence | Stable direct page results and page-level cleanup. | Suite setup/inheritance/order and exact native discovery. | | Transport | Pipe launches Perl as a FitNesse child; TCP checks a real external listener. | Same two transports with separate exact tuple expectations. |

Scenario library discrepancy

FitNesse.SuiteAcceptanceTests.SuiteSlimTests.ScenarioLibraryTestSuite.ScenariosOnlyInSlimTests is native-suite-only; it is not one of the current 50 selected individual pages. The earlier direct-page audit and TCP native baseline both recorded 2/0/0/0 (R/W/I/E) and the corrected pipe native run records 2/0/0/0. The root cause was: configure_pipe_wiki appended !define TEST_SYSTEM {slim} at wiki root, forcing generated MySuite.TestPage into Slim test selection. Pipe setup now adds only runner, command-pattern, and port defaults; TEST_SYSTEM remains page-local. Pipe and TCP native results match at 2 right and 0 wrong.

Historical direct-run versus native-suite differences

The audit also ran all 93 selected pages individually as a separate exploratory comparison. Three native-only pages had different direct and suite tuples; those direct runs are not the current 50-page regression inventory and are preserved here as separate context evidence.

Pages in both inventories

Individual inventory only

Native suite only

Property-marked pages pruned from native traversal

See the JSON manifest for every page's membership, result tuple, selection evidence, and discovery metadata.

Other wiki pages outside native results

The pinned subtree contains 136 page records stored as content.txt directories or legacy .wiki files; 43 did not produce native suite result rows.