Version 2.2.0 has a negative impact on SQLAchemy dialect's parameter handling and results

[amotl]

Problem

With version 2.2.0, released a few hours ago, we are observing a few errors with client code that uses the SQLAlchemy dialect, happening within its test suite and a few test cases in CTK.

Possibly, only a few adjustments are needed. Alternatively, we may want to roll back by pruning the release package from PyPI if the change or its impact are too serious, or not ready for GA yet.

Even after adjusting some test cases with crate/sqlalchemy-cratedb#266, doctests are failing with wrong results, see examples below.

Examples

SQLParseException[line 1:57: no viable alternative at input 'VALUES (%']

Code:
        result.rowcount
Expected:
    10
Got:
    0

Expected:
    'Algol'
Got:
    'Earth'

[amotl]

I've just yanked crate-2.2.0 from PyPI, because it showed anomalies on the SQLAlchemy dialect, as demonstrated above.

Anomalies with SQLAlchemy dialect. Better safe than sorry.