underworldcode · lmoresi · Sep 11, 2026 · Sep 11, 2026 · Sep 11, 2026
@@ -0,0 +1,90 @@
+"""Fail if a test file matches none of the globs scripts/test.sh actually runs.
+
+A batch glob that names a numeric range (``test_006[0-1]*``) stops covering the
+suite the moment someone adds ``test_0062``. That is how the whole
+integration-point suite (#703, #707) and the swarm repopulation tests (#713)
+shipped green without CI ever running them. The globs are still ranges — some
+of them have to be — so this check is what keeps them honest.
+
+Files deferred on purpose are listed in DEFERRED with the reason and the issue.
+The list may only shrink: adding to it needs a maintainer decision, exactly like
+the allowlist in check_deprecated_patterns.py.
+"""
+
+import glob
+import re
+import sys
+from pathlib import Path
+
+REPO = Path(__file__).resolve().parent.parent
+
+# Deferred by maintainer decision, not by accident.
+DEFERRED = {
+    # No issue: this band was disabled in scripts/test.sh as "potentially
+    # problematic" without one being filed. Recorded as it stands rather than
+    # dressed up — #721 follow-up work re-enables it and removes this entry.
+    "test_06[0-9][0-9]_*": "regression suite disabled in test.sh, no issue filed",
+    # Narrow: test_1072 is pulled out of this band and run by name, so a broad
+    # test_107* would list a covered file as deferred.
+    "test_1070_*": "level_2/level_3 + tier_b/tier_c, awaiting triage (#504)",
+    "test_1071_*": "level_2/level_3 + tier_b/tier_c, awaiting triage (#504)",
+    "test_1073_*": "level_2/level_3 + tier_b/tier_c, awaiting triage (#504)",
+    "test_106*": "level_2/level_3 + tier_b/tier_c, awaiting triage (#504)",
+}
+
+
+def globs_run_by(script):
+    """Every tests/... pattern passed to $PYTEST on a line that is not commented."""
+    text = script.read_text().replace("\\\n", " ")
+    patterns = set()
+    for line in text.splitlines():
+        if "$PYTEST" in line and not line.lstrip().startswith("#"):
+            patterns.update(re.findall(r"tests/[A-Za-z0-9_\[\]*.\-]+", line))
+    return patterns
+
+
+def main():
+    covered = set()
+    for pattern in globs_run_by(REPO / "scripts" / "test.sh"):
+        covered.update(glob.glob(str(REPO / pattern)))
+    covered = {Path(p).name for p in covered}
+
+    every = {p.name for p in (REPO / "tests").glob("test_*.py")}
+
+    # A deferred entry that matches nothing, or matches only files a glob
+    # already runs, is stale. Without this the list only ever grows: an entry
+    # keeps reporting a deferral that stopped being true, which is how the
+    # shrink-only invariant quietly becomes a fiction.
+    deferred, stale = set(), []
+    for pattern, reason in DEFERRED.items():
+        matched = {Path(p).name for p in glob.glob(str(REPO / "tests" / pattern))}
+        if not matched:
+            stale.append(f"{pattern!r} matches no test file")
+        elif matched <= covered:
+            stale.append(f"{pattern!r} matches only files that already run")
+        deferred.update(matched)
+
+    dark = sorted(every - covered - deferred)
+
+    if dark:
+        print(f"{len(dark)} test file(s) match no glob in scripts/test.sh:")
+        for name in dark:
+            print(f"    tests/{name}")
+        print("\nWiden the batch glob that should have caught them, or add an entry")
+        print("to DEFERRED in this file saying who deferred it and why.")
+        return 1
+
+    if stale:
+        print("DEFERRED has stale entries — the list may only shrink:")
+        for problem in stale:
+            print(f"    {problem}")
+        print("\nRemove them: the files they name are running, or no longer exist.")
+        return 1
+
+    print(f"OK: all {len(every)} test files are reachable from scripts/test.sh "
+          f"({len(deferred)} deferred by decision).")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())
@@ -0,0 +1,15 @@
+# Session scripts
+
+Drivers, plotting scripts, benchmarks and profilers written during development
+sessions. They are not tests and are not run by CI — they lived in `tests/`
+until the 2026-09 audit, where 22 of them sat among the real test files and made
+the test tree hard to read.
+
+Nothing imports these; they are run directly. Three of them
+(`vep_fault_weakening.py`, `vep_strain_weakening.py`, `vep_timedep_yield.py`)
+wrote their figures to an absolute path inside a worktree that no longer exists,
+so they had been writing nowhere since that worktree was removed; they now write
+beside themselves.
+
+A test helper belongs in `tests/` with a leading underscore (`_mg_ladder.py`),
+not here.
@@ -3,7 +3,7 @@
 against the original loop-based reference.
 
 Uses a larger mesh to make the speedup measurable.
-Run with: python tests/benchmark_index_swarm_vectorized.py
+Run with: python scripts/sessions/benchmark_index_swarm_vectorized.py
 """
 import numpy as np
 import underworld3 as uw

@@ -1,6 +1,6 @@
 """Minimal VEP timing test — isolate where time is spent.
 
-Run with: pixi run -e amr-dev python tests/minimal_vep_timing.py
+Run with: pixi run -e amr-dev python scripts/sessions/minimal_vep_timing.py
 """
 
 import time

@@ -5,7 +5,7 @@
   - Right axis: strain rate (filled grey)
 
 Reads from saved .npz data — run the benchmarks first:
-  python tests/plot_ve_oscillatory_validation.py
+  python scripts/sessions/plot_ve_oscillatory_validation.py
   python docs/advanced/benchmarks/run_ve_square_wave.py
 """
 

@@ -10,8 +10,8 @@
 Results are saved as .npz checkpoint files for re-analysis.
 
 Usage:
-    python tests/plot_ve_oscillatory_validation.py
-    python tests/plot_ve_oscillatory_validation.py --replot   # replot from saved data
+    python scripts/sessions/plot_ve_oscillatory_validation.py
+    python scripts/sessions/plot_ve_oscillatory_validation.py --replot   # replot from saved data
 """
 
 import numpy as np

@@ -3,7 +3,7 @@
 Instruments: sympy derivatives, expression unwrapping, hashing, C code generation,
 Cython compilation, and the actual PETSc solve.
 
-Run with: pixi run -e default python tests/profile_jit_phases.py
+Run with: pixi run -e default python scripts/sessions/profile_jit_phases.py
 """
 
 import time

@@ -21,7 +21,7 @@
 visual point of the figure.
 
 Run:
-    pixi run -e amr-dev python tests/run_snapshot_backstepping_demo.py
+    pixi run -e amr-dev python scripts/sessions/run_snapshot_backstepping_demo.py
 
 Output:
     snapshot_backstepping_demo.png in the current working directory.

@@ -21,7 +21,7 @@
     over-stretched, CFL violated) vs. ten substeps at sub-CFL Δt.
 
 Run:
-    pixi run -e amr-dev python tests/run_snapshot_backstepping_spatial.py
+    pixi run -e amr-dev python scripts/sessions/run_snapshot_backstepping_spatial.py
 
 Output:
     snapshot_backstepping_spatial.png in the current working directory.

@@ -1,6 +1,6 @@
 """Quick validation script for the VE shear box test.
 
-Run with: pixi run -e amr-dev python tests/run_ve_shear_validation.py
+Run with: pixi run -e amr-dev python scripts/sessions/run_ve_shear_validation.py
 """
 
 import time as timer

@@ -3,12 +3,13 @@
 Horizontal fault at y=0.5 using Surface gaussian influence function.
 Runs at two vertical resolutions to check convergence.
 
-Run: pixi run -e amr-dev python tests/vep_fault_weakening.py
+Run: pixi run -e amr-dev python scripts/sessions/vep_fault_weakening.py
 """
 
 import time
 import numpy as np
 import sympy
+import os
 import underworld3 as uw
 
 ETA = 1.0
@@ -153,6 +154,6 @@ def run_fault_model(res_x, res_y):
 fig.suptitle(f"VEP embedded fault convergence: $\\tau_y$={TAU_Y_FAULT}/{TAU_Y_BULK}, width={FAULT_WIDTH}",
              fontsize=12, y=1.02)
 fig.tight_layout()
-out_path = "/Users/lmoresi/+Underworld/underworld3-pixi/.claude/worktrees/solver-unification/vep_fault.png"
+out_path = os.path.join(os.path.dirname(__file__), "vep_fault.png")
 fig.savefig(out_path, dpi=150, bbox_inches='tight')
 uw.pprint(0, f"Saved {out_path}")
@@ -8,12 +8,13 @@
 
 No evaluate, no projection — pure numpy on stored data.
 
-Run: pixi run -e amr-dev python tests/vep_strain_weakening.py
+Run: pixi run -e amr-dev python scripts/sessions/vep_strain_weakening.py
 """
 
 import time
 import numpy as np
 import sympy
+import os
 import underworld3 as uw
 
 ETA = 1.0
@@ -145,6 +146,6 @@
 axes[2].grid(True, alpha=0.3)
 
 fig.tight_layout()
-out_path = "/Users/lmoresi/+Underworld/underworld3-pixi/.claude/worktrees/solver-unification/vep_strain_weakening.png"
+out_path = os.path.join(os.path.dirname(__file__), "vep_strain_weakening.png")
 fig.savefig(out_path, dpi=150)
 print(f"Saved {out_path}")
@@ -6,12 +6,13 @@
 No strain accumulation, no projection of viscosity ratios —
 just the solver responding to a changing yield stress parameter.
 
-Run: pixi run -e amr-dev python tests/vep_timedep_yield.py
+Run: pixi run -e amr-dev python scripts/sessions/vep_timedep_yield.py
 """
 
 import time
 import numpy as np
 import sympy
+import os
 import underworld3 as uw
 
 ETA = 1.0
@@ -132,6 +133,6 @@
 ax.grid(True, alpha=0.3)
 
 fig.tight_layout()
-out_path = "/Users/lmoresi/+Underworld/underworld3-pixi/.claude/worktrees/solver-unification/vep_timedep_yield.png"
+out_path = os.path.join(os.path.dirname(__file__), "vep_timedep_yield.png")
 fig.savefig(out_path, dpi=150)
 print(f"Saved {out_path}")
@@ -91,6 +91,12 @@ if [ $PARALLEL_ONLY -eq 0 ]; then
   echo "Running serial test suite..."
   echo ""
 
+  # Every test file must be reachable from a glob below. #721 was a range that
+  # stopped covering its own band; 586224f6 took 006x-007x whole so that one
+  # cannot recur, but the other ranges are still hand-maintained. This is what
+  # notices when one of them goes stale.
+  python3 "$(dirname "$0")/check_test_coverage.py" || status=1
+
   # Run simple tests (0000-0299: basic functionality, imports, simple operations)
   $PYTEST tests/test_00[0-4]*py || status=1
   $PYTEST tests/test_0050*py || status=1
@@ -102,7 +108,7 @@ if [ $PARALLEL_ONLY -eq 0 ]; then
   # siblings is not dark again. Verified passing (103 tests) before wiring in.
   $PYTEST tests/test_005[1-9]*py tests/test_00[6-7]*py || status=1
   $PYTEST tests/test_01*py || status=1
-  $PYTEST tests/test_02*py || status=1
+  $PYTEST tests/test_02*py tests/test_03*py || status=1
 
   # Intermediate tests (0500-0799: data structures, transformations, enhanced interfaces)
   # NOTE: Temporarily disabling test_06*py regression tests (potentially problematic)
@@ -113,7 +119,7 @@ if [ $PARALLEL_ONLY -eq 0 ]; then
   $PYTEST tests/test_08*py || status=1
 
   # Poisson solvers (including Darcy flow)
-  $PYTEST tests/test_100[0-9]*py || status=1
+  $PYTEST tests/test_100[0-9]*py tests/test_103*py || status=1
 
   # Solver / system tests (advanced solver problems)
   # test_101* / test_102* include the rotated free-slip suite (test_1018,