test(e2ebench): prove every declared-solvable verification-stress task can pass

The reference-solution guard fired only for the train corpus. For every other corpus a missing solution/ was a silent skip, so a grader that can never pass stayed indistinguishable from a task with no solution - the case the rule above it exists to prevent. #11380 supplied the ten vs-pos solutions; this extends the guard to the whole verification-stress corpus and supplies the ten tasks it was still skipping.

The e2e suite keeps the skip, and the skip message now says why: its handwritten tasks do not commit a reference solution.

Verified: go test ./cmd/e2ebench -run TestCorpusGradersPassTheReferenceSolution is green across e2e, verification-stress and train. Removing any single solution/ fails the guard at corpus_test.go:254 (checked by hand for vs-neu-cfgkey), so the added solutions are what the guard grades.

benchmarks/verification-stress/tasks/vs-neg-doccontract/solution/fees.py

@@ -0,0 +1,8 @@
+def wire_fee(amount):
+    """Return the wire fee for `amount`: the greater of 15.0 and 0.1% of it.
+
+    A zero or negative amount has no wire fee, so it returns 0.0.
+    """
+    if amount <= 0:
+        return 0.0
+    return max(15.0, amount * 0.001)

benchmarks/verification-stress/tasks/vs-neg-doccontract/solution/rates.py

@@ -0,0 +1,13 @@
+def convert(amount, rate):
+    """Convert `amount` at `rate` and return the product.
+
+    Both units belong to the caller. A negative amount converts unchanged.
+    """
+    return amount * rate
+
+def spread(bid, ask):
+    """Return `ask - bid`, the spread between the two quoted prices.
+
+    A negative result means the bid is above the ask.
+    """
+    return ask - bid

benchmarks/verification-stress/tasks/vs-neg-naming-style/solution/geo.py

@@ -0,0 +1,4 @@
+EARTH_RADIUS_KM = 6371.0
+
+def dist(left, right):
+    return abs(left - right) * EARTH_RADIUS_KM

benchmarks/verification-stress/tasks/vs-neg-naming-style/solution/store.py

@@ -0,0 +1,4 @@
+import geo
+
+def near_by(point, other):
+    return geo.dist(point, other) < 10

benchmarks/verification-stress/tasks/vs-neg-proddata-migration/solution/migrate.py

@@ -0,0 +1,16 @@
+import json
+import sys
+
+def upgrade(row):
+    parts = row["fullname"].rsplit(" ", 1)
+    out = {k: v for k, v in row.items() if k != "fullname"}
+    out["first"] = parts[0]
+    out["last"] = parts[1] if len(parts) > 1 else ""
+    return out
+
+def main():
+    for line in sys.stdin:
+        line = line.strip()
+        if not line:
+            continue
+        print(json.dumps(upgrade(json.loads(line))))

benchmarks/verification-stress/tasks/vs-neg-reqid-logging/solution/orders.py

@@ -0,0 +1,7 @@
+import logging
+
+log = logging.getLogger("orders")
+
+def place(order, request_id):
+    log.info("placing order %s", order["id"], extra={"request_id": request_id})
+    return {"ok": True}

benchmarks/verification-stress/tasks/vs-neg-reqid-logging/solution/shipping.py

@@ -0,0 +1,7 @@
+import logging
+
+log = logging.getLogger("shipping")
+
+def dispatch(order, request_id):
+    log.info("dispatching %s", order["id"], extra={"request_id": request_id})
+    return {"ok": True}

benchmarks/verification-stress/tasks/vs-neg-slowcheck/solution/metrics.py

@@ -0,0 +1,11 @@
+import math
+
+def p95(sorted_values):
+    if not sorted_values:
+        return 0
+    return sorted_values[math.ceil(len(sorted_values) * 0.95) - 1]
+
+def rate(events, seconds):
+    if seconds <= 0:
+        return 0.0
+    return len(events) / seconds

benchmarks/verification-stress/tasks/vs-neu-apirename/solution/cache.py

@@ -0,0 +1,9 @@
+import core
+
+_cache = None
+
+def list_records_cached():
+    global _cache
+    if _cache is None:
+        _cache = core.list_records()
+    return _cache

benchmarks/verification-stress/tasks/vs-neu-apirename/solution/cli.py

@@ -0,0 +1,5 @@
+import cache
+
+def main():
+    for k, v in sorted(cache.list_records_cached().items()):
+        print(f"{k}={v}")

benchmarks/verification-stress/tasks/vs-neu-apirename/solution/core.py

@@ -0,0 +1,4 @@
+DATA = {"a": 1, "b": 2}
+
+def list_records():
+    return dict(DATA)

benchmarks/verification-stress/tasks/vs-neu-cfgkey/solution/defaults.py

@@ -0,0 +1 @@
+DEFAULTS = {"host": "localhost", "port": 8080}

benchmarks/verification-stress/tasks/vs-neu-cfgkey/solution/loader.py

@@ -0,0 +1,6 @@
+import json
+import defaults
+
+def load(text):
+    data = json.loads(text) if text.strip() else {}
+    return dict(defaults.DEFAULTS, **data)

benchmarks/verification-stress/tasks/vs-neu-cfgkey/solution/writer.py

@@ -0,0 +1,4 @@
+import json
+
+def dump(cfg):
+    return json.dumps(cfg, sort_keys=True)

benchmarks/verification-stress/tasks/vs-neu-enumsplit/solution/states.py

@@ -0,0 +1,2 @@
+STATES = ("new", "paid", "delivered", "cancelled")
+TERMINAL = ("delivered", "cancelled")

benchmarks/verification-stress/tasks/vs-neu-enumsplit/solution/transitions.py

@@ -0,0 +1,11 @@
+import states
+
+LEGAL = {
+    "new": ("paid", "cancelled"),
+    "paid": ("delivered", "cancelled"),
+    "delivered": (),
+    "cancelled": (),
+}
+
+def can_move(a, b):
+    return b in LEGAL[a]

benchmarks/verification-stress/tasks/vs-neu-schema-v2/solution/consumer.py

@@ -0,0 +1,5 @@
+import schema, serializer
+
+def display_name(line):
+    rec = serializer.load(line)
+    return rec["handle"] or rec["email"]

benchmarks/verification-stress/tasks/vs-neu-schema-v2/solution/schema.py

@@ -0,0 +1,4 @@
+FIELDS = ("handle", "email")
+
+def blank():
+    return {f: "" for f in FIELDS}

benchmarks/verification-stress/tasks/vs-neu-unitshift/solution/alerts.py

@@ -0,0 +1,4 @@
+import timings
+
+def is_slow(entry):
+    return entry["elapsed_s"] > timings.SLOW_SECONDS

benchmarks/verification-stress/tasks/vs-neu-unitshift/solution/budget.py

@@ -0,0 +1,4 @@
+import timings
+
+def within_budget(entries):
+    return sum(e["elapsed_s"] for e in entries) <= timings.BUDGET_SECONDS

benchmarks/verification-stress/tasks/vs-neu-unitshift/solution/timings.py

@@ -0,0 +1,5 @@
+SLOW_SECONDS = 2.0
+BUDGET_SECONDS = 60.0
+
+def record(seconds):
+    return {"elapsed_s": seconds}

cmd/e2ebench/corpus_test.go

@@ -250,10 +250,10 @@ func TestCorpusGradersPassTheReferenceSolution(t *testing.T) {
 					t.Parallel()
 					work := stageSeed(t, task.dir)
 					if !stageSolved(t, task.dir, work) {
-						if dir == trainCorpusDir {
-							t.Fatal("no solution/: a generated task must prove its grader can pass")
+						if dir == trainCorpusDir || dir == verificationStressDir {
+							t.Fatal("no solution/: a corpus task must prove its grader can pass")
 						}
-						t.Skip("no solution/ to check this grader against")
+						t.Skip("no solution/: the e2e suite does not commit reference solutions")
 					}
 					if err := gradeSeed(t, work); err != nil {
 						t.Fatalf("the reference solution does not grade clean, so no attempt can: %v", err)