Test Execution Duration Report Modernized

Test Execution Duration & Throughput report (modernized)

Tracking issue: #1975 (parent metric: #1073, Metrics & Trends — average test execution duration / execution rate).

What it is

A net-new TestLink 2.0.1 report under Reports. TestLink 1.9.20 stores the actual duration of every execution in executions.execution_duration (minutes, decimal(6,2)) but ships no screen that reports it: the execution history of a test case shows one run at a time, and every 1.9.20 report aggregates status only. As a result a team could not answer:

• how long execution takes for a plan,
• which test cases are the slowest,
• how many executions per day were recorded.

This screen answers those questions for a rolling window.

Screen

gui/templates/results/execDuration.html

• Dashio teal header, dark toolbar (Refresh, Export CSV, Build, Platform, Window, Slow threshold, Apply, Close).
• Report-context panel (project, plan, build, platform, window, threshold).
• 10 KPI cards: executions, executions with duration, duration coverage %, average, median, P90, total minutes, throughput per day, slow versions, distinct testers.
• Verdict banner (no executions / slow versions / all normal / no duration recorded).
• Per-build breakdown table (executions, timed executions, total and average minutes).
• Main 13-column DataTable, slowest average first: test case, external id, suite, verdict, executions, with duration, avg/min/max/total minutes, last status, last run, builds.
• CSV export (BOM + spreadsheet-formula neutralisation).
• Localized state cards for 400 / 401 / 403 / 404 / 405.

BFF

api/execduration/index.php — plain PHP, session auth, bffSameOriginGuard, bffEnforceSession, JSON I/O, Cache-Control: no-store.

Action	Method	Parameters
init	GET|HEAD	tplan_id, optional tproject_id
report	GET|HEAD	tplan_id, optional tproject_id, build_id, platform_id, window_days (1–365, default 30), slow_min (1–1440, default 30)

Rights: testplan_metrics on the owning test project of the plan (getAccess = true). The client-supplied tproject_id is only an assertion and can never widen the query.

Error contract: not_authenticated (401), no_right (403), tplan_not_found / project_mismatch (404), invalid_request / method_not_allowed (400/405).

Semantics

• The analysed window is the last window_days days ending now; only executions with an execution_ts inside the window are counted, so the throughput figure is meaningful.
• execution_duration is the actual duration in minutes. A NULL or 0 duration means "not recorded" and is excluded from the duration maths but still counts for throughput and status distribution.
• The unit of the table is the test-case version (tcversion_id); only versions currently linked to the plan (testplan_tcversions) are considered, so an execution that survives an un-assignment cannot resurrect a version.
• A version is slow when its average recorded duration is at least slow_min minutes.

Wiring

• cfg/reports.cfg.php: $tlCfg->reports_list['exec_duration'].
• api/aside/index.php: Reports menu mapping for link_report_exec_duration.
• api/resultsnav/index.php: exec_duration → modern screen.
• lib/functions/common.php: $actions->execDuration.
• locale/*/strings.txt: $TLS_link_report_exec_duration (19 server locales).
• i18n: execDuration.* keys + footers.execDuration in all 10 bundles.

Verification

• Fixture: tmp/fixtures_1953.php (plan 4) with execution durations seeded via SQL (execution_duration = id * 15, execution_ts = NOW() - id days).
• BFF: init 200; report 200 with 7 executions / 420 min / avg 60 / median 60 / P90 105; build filter narrows to 1 execution; no session 401; unknown plan 404; project mismatch 404; invalid window 400; no-right user 403.
• Browser (admin, EN): report renders, filters apply, CSV button enables, console clean, no raw i18n keys.
• i18n coverage gate: 9/9 bundles PASS (10228 keys).
• Suite: Issue #1975 in tmp/TLU_Test_Cases.md = 12/12 PASS.

Screenshots