Toby Mao

Currently when a job is canceled (sigint, sigterm), retries is incremented. I'd like for there to be an option for these types of failures to not increment retries. Another way...

add getitem notation

1

presto, hive, spark all use bracket notation for map access field['foo'] etc

[FEATURE] example usage of sqlglot to simplify code

2

obviously there'd be a lot more work, but i wanted to kind of demo how much easier it will be to do things

get_column_in_relation currently returns a base column class https://github.com/TobikoData/sqlmesh/pull/1873/files we had to revert the change to the engine specific subclasses because each subclass has a different init signature. from_description crashes on...

Add notify_user to airflow

1

Airflow doesn't use the scheduler so doesn't alert on audit failures. Either engines/commands.py needs to start doing that, or perhaps we move the notification into the snapshot_evaluator directly so that...

although we don't recommend or document this, we can expose this for certain advanced users as an escape hatch.