Hacktoberfest 2026: los issues que los mantenedores marcaron para octubre, abiertos y aptos para principiantes. Explorar issues de Hacktoberfest

do_rollback() swallows the connector's NotSupportedError, so a rollback reports success over a write that stands

Abierto Apto para principiantes
  • ¿Algo no está bien?
#77 0 comentarios 0 reacciones 0 asignados Ver en GitHub

@dwaddle ya está trabajando en esto.

Desde el 23/10/2026.

  • #86 de @dwaddle — abierto

Evaluación

Dificultad
2/5
Tiempo estimado
1-3 horas
Aptitud para principiantes
68/100
Tipo de issue
Error
Claridad
Bastante claro
Estado de actividad
Tranquilo
Stack tecnológico
python, sqlalchemy
Área
databases

Línea de trabajo

Localice DatabendDialect.do_rollback e inspeccione el comportamiento de connector Connection.rollback descrito en el issue. Verifique que rollback ya no informe de éxito ante una operación no compatible, y añada o actualice la cobertura para las rutas de error de rollback explícito y SQLAlchemy si el repositorio tiene pruebas correspondientes.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

Versions

  • databend-sqlalchemy 0.5.5
  • databend-driver 0.34.2
  • SQLAlchemy 2.0.51
  • Python 3.12.9, macOS
  • Server: datafuselabs/databend:v1.2.925-patch-4 in Docker

What happens

connector.Connection.rollback() is honest — it raises:

class Connection:
    def commit(self): pass
    def rollback(self):
        raise NotSupportedError("Transactions are not supported")

but the dialect overrides do_rollback and swallows it:

class DatabendDialect:
    def do_rollback(self, dbapi_connection):
        # No transactions
        pass

So sqlalchemy.Connection.rollback() returns normally over a write that is still there afterwards. Nothing raises, nothing warns.

Three sequences, one variable changed each time, row count read back on a fresh connection:

sequence rows afterwards
INSERT, connection closed without committing 1
INSERT, then an explicit Connection.rollback() 1
BEGIN, INSERT, ROLLBACK sent as plain statements 1

Expected: either the rollback raises (the connector underneath already does), or it rolls something back.

Why this is worse than simply not having transactions

Wrapping writes in a transaction and rolling back on error is ordinary code, and this turns it from loudly unsupported into silently wrong. with engine.begin(): around a block that fails reports a clean rollback and leaves the partial write in place. SQLAlchemy's own error path issues a rollback when a statement fails inside a transaction, so that gets the same false report without the caller writing a rollback at all.

I ran into it building a compatibility layer that treats "can this engine undo a write" as a capability it has to answer per backend. Databend is the second engine I've seen answer that question wrongly, and the only one where the wrong answer comes from a deliberate override rather than an accident further down.

One thing worth stating precisely

"Databend has no transactions" isn't accurate, and the distinction matters for how you'd fix this. BEGIN/COMMIT are real statements, and every HTTP response carries a txn_state field (AutoCommit, in this configuration). Reaching them needs the client to carry the session back between requests — session_id and need_sticky in the response are how — and the blocking cursor path in databend-driver doesn't. So transactions are unreachable through this adapter rather than absent from the database. That's clearly a bigger piece of work; this issue is only about the false success.

Related: get_isolation_level raises NotImplementedError, so a caller can't ask the engine what it's going to do instead.

Suggested fix

Drop the do_rollback override and let the connector's NotSupportedError propagate. If that breaks too much for existing users, a warning would at least make it discoverable. The current shape is the one that can't be detected from outside.

Lenguaje dominante
Python
Estrellas
15
Forks
5
Merge medio
1 d 15 h
PR fusionados (30 d)
1

Preparar el entorno

Este proyecto no incluye contenedor de desarrollo, Dockerfile ni guía de contribución, así que la configuración corre por tu cuenta: empieza por su README y consulta nuestra guía para la primera contribución para los pasos generales.

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de databendlabs/databend-sqlalchemy

Todos los issues de databendlabs/databend-sqlalchemy

Issues similares

Más issues de Python

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.