FA-97881 / Dataset split isolation / Open access

Dataset split: session side lookup uses the date instead of the session · case 01

session side lookup uses the date instead of the session.

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

ROOT CAUSE

The implementation looks up each row’s date in a map whose keys are session IDs, so the session’s assigned side is ignored.

THE FAILURE

The implementation looks up each row’s date in a map whose keys are session IDs, so the session’s assigned side is ignored.

Unsuccessful approach: Using a date as the ownership key would still split one session across dates.

Case contract

Rows [id,session,date]. Return IDs belonging to sessions mapped to train; dates do not affect ownership.

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(rows, sides):
    return sorted(i for i,s,d in rows if sides.get(d)=='train')
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('cross date session',solve([['a','s','d1'],['b','s','d2']],{'s':'train'}),['a','b'])
check('validation session',solve([['a','v','d1']],{'v':'validation'}),[])
check('unknown session',solve([['a','u','train']],{}),[])
check('empty',solve([],{}),[])
check('date cannot override',solve([['a','s','holdout']],{'s':'train'}),['a'])
check('multiple sessions',solve([['a','s','x'],['b','t','x']],{'s':'train','t':'validation'}),['a'])
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
cross date session	`[]`	`['a', 'b']`	Failed
validation session	`[]`	`[]`	Passed
unknown session	`[]`	`[]`	Passed
empty	`[]`	`[]`	Passed
date cannot override	`[]`	`['a']`	Failed
multiple sessions	`[]`	`['a']`	Failed

SHA-256 / 68f44fd22fb55c525618b7647666c2e39dbcce5c7d65ca5c43827d347bef6114

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(rows, sides):
    return sorted(i for i,s,d in rows if sides.get(s,'validation')=='train' and d!='holdout')
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('cross date session',solve([['a','s','d1'],['b','s','d2']],{'s':'train'}),['a','b'])
check('validation session',solve([['a','v','d1']],{'v':'validation'}),[])
check('unknown session',solve([['a','u','train']],{}),[])
check('empty',solve([],{}),[])
check('date cannot override',solve([['a','s','holdout']],{'s':'train'}),['a'])
check('multiple sessions',solve([['a','s','x'],['b','t','x']],{'s':'train','t':'validation'}),['a'])
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
cross date session	`['a', 'b']`	`['a', 'b']`	Passed
validation session	`[]`	`[]`	Passed
unknown session	`[]`	`[]`	Passed
empty	`[]`	`[]`	Passed
date cannot override	`[]`	`['a']`	Failed
multiple sessions	`['a']`	`['a']`	Passed

SHA-256 / 56b68435bc3764aed3005b450a156e9709a797540584036dbffcaa51eba9dccd

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 2026-09-30T16:48:07.117919+00:00.

Case digest / 4e350dd048ffb10a39309c0de1c69c75f3822a49c2e7ea8d50cc43f9c0f79863