Hacktoberfest 2026 : les issues que les mainteneurs ont marquées pour octobre, ouvertes et accessibles aux débutants. Parcourir les issues Hacktoberfest

Percentage split and variant hash collide when segment key equals feature key

Ouverte
#348 0 commentaires 0 réactions 0 personnes assignées Voir sur GitHub

Personne n'a encore pris cette issue.

Évaluation

Difficulté
3/5
Temps estimé
1-2 jours
Accessibilité débutants
78/100
Type d'issue
Bug
Clarté
Clairement spécifiée
Activité
Active
Stack technique
python
Domaine
backend

Piste de recherche

Start in segments/evaluator.py at the PERCENTAGE_SPLIT handling and variant selection, then trace the get_evaluation_result entry point used by the reproduction. Verify that percentage splits and multivariate variants no longer collide when their keys match, and rerun the 1,000-identity example to confirm both variants are served.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

Bug

A percentage split segment and a multivariate feature hash the same inputs when the segment key equals the feature key:

  • the split hashes [segment_key, identity_key] (segments/evaluator.py, PERCENTAGE_SPLIT)
  • the variant hashes [feature_key, identity_key] (segments/evaluator.py, variant selection)

Both give the same number. An identity is only in a 50% split when that number is 50 or below, so it always lands in the first variant. The other variants are never served to anyone in the segment.

Segment and feature keys are ids from different tables, so they can be equal by chance.

Reproduction

from flag_engine.utils.hashing import get_hashed_percentage_for_object_ids as h

identity = "user-1"
assert h(["95", identity]) == h(["95", identity])  # split == variant

End to end: a segment with a 50% split on $.identity.key, overriding a feature with two 50/50 variants, evaluated for 1000 identities:

from collections import Counter

from flag_engine.engine import get_evaluation_result


def variants(segment_key: str, feature_key: str) -> Counter[str]:
    context = {
        "environment": {"key": "env", "name": "env"},
        "segments": {
            segment_key: {
                "key": segment_key,
                "name": "rollout",
                "rules": [{"type": "ALL", "conditions": [
                    {"property": "$.identity.key", "operator": "PERCENTAGE_SPLIT", "value": "50"},
                ]}],
                "overrides": [{
                    "key": feature_key, "name": "checkout", "enabled": True, "value": "control",
                    "variants": [
                        {"value": "a", "weight": 50, "priority": 0, "key": "variant-a"},
                        {"value": "b", "weight": 50, "priority": 1, "key": "variant-b"},
                    ],
                }],
            }
        },
        "features": {"checkout": {"key": "1", "name": "checkout", "enabled": False, "value": None}},
    }
    counts: Counter[str] = Counter()
    for i in range(1000):
        context["identity"] = {"identifier": f"user-{i}", "key": f"user-{i}"}
        flag = get_evaluation_result(context)["flags"]["checkout"]
        if flag["enabled"]:
            counts[flag["variant"]] += 1
    return counts


print(variants("95", "96"))  # Counter({'variant-b': 260, 'variant-a': 254})
print(variants("95", "95"))  # Counter({'variant-a': 514})

Impact

Any segment override with variants on a percentage split segment whose key equals the feature key serves only one variant to the whole segment. For an experiment rollout, the experiment has a single arm. Seen in Flagsmith's API tests in Flagsmith/flagsmith#8646.

Langage dominant
Python
Étoiles
10
Forks
5
Merge moyen
7 j 22 h
PR mergées (30 j)
1

Préparer son environnement

Ce projet ne fournit ni conteneur de développement, ni Dockerfile, ni guide de contribution : l'installation est à votre charge. Commencez par son README, et consultez notre guide de la première contribution pour les étapes générales.

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de Flagsmith/flagsmith-engine

Toutes les issues de Flagsmith/flagsmith-engine

Issues similaires

Plus d'issues Python

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.