FA-97926 / Dataset split isolation / Open access

Dataset split: IDs from later partition manifests disappear · case 01

IDs from later partition manifests disappear.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

The implementation returns only the first partition’s IDs, discarding all subsequent membership lists.

THE FAILURE

The implementation returns only the first partition’s IDs, discarding all subsequent membership lists.

Unsuccessful approach: Removing only adjacent duplicates still repeats an ID that reappears after another ID.

Case contract

Partitions map names to lists of string IDs; return the flattened IDs in partition/input order, with duplicates removed globally.

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(partitions):
    return list(dict.fromkeys(next(iter(partitions.values()),[])))
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('overlap assigned once',solve({'train':['a','b'],'validation':['b','c']}),['a','b','c'])
check('empty',solve({}),[])
check('empty partition',solve({'train':[]}),[])
check('within duplicate',solve({'train':['a','a']}),['a'])
check('order stable',solve({'validation':['z'],'train':['a','z']}),['z','a'])
check('three sides',solve({'a':['x'],'b':['y'],'c':['x','z']}),['x','y','z'])
check('nonadjacent duplicate',solve({'train':['a','b'],'validation':['a']}),['a','b'])
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
overlap assigned once	`['a', 'b']`	`['a', 'b', 'c']`	Failed
empty	`[]`	`[]`	Passed
empty partition	`[]`	`[]`	Passed
within duplicate	`['a']`	`['a']`	Passed
order stable	`['z']`	`['z', 'a']`	Failed
three sides	`['x']`	`['x', 'y', 'z']`	Failed
nonadjacent duplicate	`['a', 'b']`	`['a', 'b']`	Passed

SHA-256 / 8faa883e9b92010388ccdd65a5554ddd1cede83e6acff46b193a34823020810a

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(partitions):
    out=[]
    for x in (x for values in partitions.values() for x in values):
     if not out or out[-1]!=x: out.append(x)
    return out
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('overlap assigned once',solve({'train':['a','b'],'validation':['b','c']}),['a','b','c'])
check('empty',solve({}),[])
check('empty partition',solve({'train':[]}),[])
check('within duplicate',solve({'train':['a','a']}),['a'])
check('order stable',solve({'validation':['z'],'train':['a','z']}),['z','a'])
check('three sides',solve({'a':['x'],'b':['y'],'c':['x','z']}),['x','y','z'])
check('nonadjacent duplicate',solve({'train':['a','b'],'validation':['a']}),['a','b'])
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
overlap assigned once	`['a', 'b', 'c']`	`['a', 'b', 'c']`	Passed
empty	`[]`	`[]`	Passed
empty partition	`[]`	`[]`	Passed
within duplicate	`['a']`	`['a']`	Passed
order stable	`['z', 'a', 'z']`	`['z', 'a']`	Failed
three sides	`['x', 'y', 'x', 'z']`	`['x', 'y', 'z']`	Failed
nonadjacent duplicate	`['a', 'b', 'a']`	`['a', 'b']`	Failed

SHA-256 / 5026040b9f5a5790758f7df0e0ed8dc4bd6003d4012e688fba4dd3c9a1bf0c82

HELD IN THE MEMBER ARCHIVE

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

This mechanism has 7 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 2026-09-30T16:48:07.548583+00:00.

Case digest / bee5afccda17cd215d694eb2ea0e8c7cdde89f7f09aae02eafbaa78888f9613d