make the default behavior for proto writing not-backwards compatible
- Langage dominant
- Java
- Étoiles
- 3.1k
- Forks
- 1.6k
- Merge moyen
- 3 j 12 h
- PR mergées (30 j)
- 33
Description
https://issues.apache.org/jira/browse/PARQUET-968 introduced supporting maps and lists in a spec compliant way. however, to not break existing libraries, a flag was introduced and defaulted the write behavior to NOT use the specs compliant writes.
it's been over 5 years, and people should be really off of it. so much so, that trying to use the new parquet-cli tool to read parquet files generated by flink doesn't work b/c it's hard coded to never allow the old style. the deprecated parquet-tools reads these files fine b/c it's the older style.
i started coding up a workaround in flink-parquet and parquet-cli, but stopped. we really should just move on at this point, imho. protobufs often have repeated primitives and maps, so it's more pressing to get proper specs compliant support for it now. we should keep the flag around and let people override it back to being backwards compatible though.
i have the code written and can submit a PR if you'd like.
i'm not an expert in parquet though, so i'm unclear as to the deep downstream ramifications of this change, so i would love to get feedback in this area.
**Reporter**: [J Y](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=jinyius) / @jinyius
**Note**: *This issue was originally created as [PARQUET-2180](https://issues.apache.org/jira/browse/PARQUET-2180). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*
Guide de contribution
Aucun guide de contribution indexé pour ce dépôt
Piste de recherche
Examinez d’abord PARQUET-968 et la documentation de migration, puis inspectez la gestion existante des flags dans flink-parquet et parquet-cli. Vérifiez que la valeur par défaut devienne conforme à la spécification tout en laissant disponible la surcharge rétrocompatible, et testez parquet-cli avec des fichiers générés par Flink et avec l’ancien style.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Évaluation
- Stack technique
- java
- Domaine
- data-engineering
- Type d'issue
- Fonctionnalité
- Difficulté
- 4/5
- Temps estimé
- 3-5 jours
- Activité
- À l'abandon
- Clarté
- Plutôt claire
- Accessibilité débutants
- 35/100