FA-97861 / Dataset split isolation / Open access

Dataset split: validation begins inside the embargo window · case 01

validation begins inside the embargo window.

Verified by executionVariant 1 · 6 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

A temporal holdout leaves only a strict ordering constraint, not a safety gap.

THE FAILURE

A temporal holdout leaves only a strict ordering constraint, not a safety gap.

Unsuccessful approach: A one-tick gap still leaks when embargo spans multiple ticks.

Case contract

Return IDs with time < start-gap from [id,time]; gap is a nonnegative integer and start is exclusive.

Why this case matters

This toy split allocator makes membership rules explicit for reproducible evaluation fixtures.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(records, start, gap):
    return sorted(i for i,t in records if t<start)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('outside gap',solve([['a',N-4]],N,3),['a'])
check('edge excluded',solve([['a',N-3]],N,3),[])
check('inside excluded',solve([['a',N-1]],N,3),[])
check('start excluded',solve([['a',N]],N,3),[])
check('zero gap',solve([['a',N-1]],N,0),['a'])
check('empty',solve([],N,2),[])
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
outside gap	`['a']`	`['a']`	Passed
edge excluded	`['a']`	`[]`	Failed
inside excluded	`['a']`	`[]`	Failed
start excluded	`[]`	`[]`	Passed
zero gap	`['a']`	`['a']`	Passed
empty	`[]`	`[]`	Passed

SHA-256 / 86c9e586450ef7e4d5db0164d0a9b5d322c9a9e62dac02414c808fcfccd406c2

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(records, start, gap):
    return sorted(i for i,t in records if t<start-1)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('outside gap',solve([['a',N-4]],N,3),['a'])
check('edge excluded',solve([['a',N-3]],N,3),[])
check('inside excluded',solve([['a',N-1]],N,3),[])
check('start excluded',solve([['a',N]],N,3),[])
check('zero gap',solve([['a',N-1]],N,0),['a'])
check('empty',solve([],N,2),[])
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
outside gap	`['a']`	`['a']`	Passed
edge excluded	`['a']`	`[]`	Failed
inside excluded	`[]`	`[]`	Passed
start excluded	`[]`	`[]`	Passed
zero gap	`[]`	`['a']`	Failed
empty	`[]`	`[]`	Passed

SHA-256 / b5694628bf54be0903384b8f33c7ed7635f8b28bb84ea5bb6b6e5dfc8b525867

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

Inputs are small in-memory records with self-contained contracts; no claim is made about production sampling quality. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.

Observations recorded using Python 3.12.14 at 2026-09-30T16:48:06.688142+00:00.

Case digest / f2981c0b80f5e59c20ee3eed7f2ec71b6fac2a724fa0de45a8c7778bf6a71fd8