Tests fail with SQLAlchemy 1.4

[UjinT34]

Tests fail with SQLAlchemy 1.4:

ERROR collecting graphene_sqlalchemy/tests/test_converter.py
ImportError while importing test module 'graphene-sqlalchemy/graphene_sqlalchemy/tests/test_converter.py'.                                                                                                                    
Hint: make sure your test modules/packages have valid Python names.                                                               
Traceback:                                                                                                                        
graphene_sqlalchemy/tests/test_converter.py:9: in <module>                                                                        
    from sqlalchemy_utils import ChoiceType, JSONType, ScalarListType                                                             
.tox/py39/lib/python3.9/site-packages/sqlalchemy_utils/__init__.py:1: in <module>                                                 
    from .aggregates import aggregated  # noqa                                                                                    
.tox/py39/lib/python3.9/site-packages/sqlalchemy_utils/aggregates.py:372: in <module>                                             
    from .functions.orm import get_column_key
.tox/py39/lib/python3.9/site-packages/sqlalchemy_utils/functions/__init__.py:1: in <module>
    from .database import (  # noqa
.tox/py39/lib/python3.9/site-packages/sqlalchemy_utils/functions/database.py:10: in <module>
    from ..expressions import explain_analyze
.tox/py39/lib/python3.9/site-packages/sqlalchemy_utils/expressions.py:4: in <module>
    from sqlalchemy.sql.expression import (
E   ImportError: cannot import name '_literal_as_text' from 'sqlalchemy.sql.expression' (graphene-sqlalchemy/.tox/py39/lib/python3.9/site-packages/sqlalchemy/sql/expression.py)

sqlalchemy/sqlalchemy@f07e050

The library itself seems to work fine with batching disabled.

[tad3j]

I've noticed the same with batching. I think the reason is that they changed QueryContext in SQLAlchemy v1.4:

v1.3 usage (used by this lib):

graphene-sqlalchemy/graphene_sqlalchemy/batching.py

Line 55 in 20ecaea

query_context = QueryContext(session.query(parent_mapper.entity))

In v1.4 QueryContext expects multiple arguments while v1.3 only had a single one: https://github.com/sqlalchemy/sqlalchemy/blob/cb8906bab1776bafed48ef69ded0768461f7e7b8/lib/sqlalchemy/orm/context.py#L50

Any plans to update batching to work with 1.4? Without that we should lock our SQLALchemy versions to v1.3 if we want to use batching.

If support is added, should it support both v1.3 and v1.4 QueryContext?

EDIT: This is the exception I get when querying:

[zsiciarz]

This has been addressed in #317 .

[raguiar2]

Is a release going to be cut with this feature? It would be useful to have the most up-to-date version of graphene-sqlalchemy on pypi so it can be used with 1.4

[HansG89]

What about this issue? Still same behavior in latest version. Batching with sqlalchemy 1.4 not working

[erikwrede]

The graphene-sqlalchemy v3 beta release contains this fix. Since there's a lot of changes in the graphene-v3 PR, additional work would have to be accomplished to fix this in 2.3.0. Would this be required, or are there any additional blockers for upgrading to 3.0?

[jaredculp]

Bump on the question above. Could this fix be released to the 2.x branch?

[erikwrede]

@jaredculp What is your blocker for upgrading to graphene 3 (apart from the SQLAlchemy beta status, which is now actively being brought towards release)?

I'm asking because focusing on backporting will reduce my time for working on the 3.0 release, so unless there's a compelling reason, I sadly cannot focus on more versions than 3.0.
But as stated before, I'd welcome and support any efforts on backporting this.

Possible tasks for that would be

• Create a 2.x legacy branch from the release tag
• Untangle commit history from master, separating 3.0 changes from this fix
• Push to the new branch and ensure tests & ci & prs are running

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related topics referencing this issue.