There has been some work on an SQL library which is experimental work. Currently, there are multiple versions and branches, see below for some references.
Context:
This is a very large issue, which needs to be broken down into smaller tasks at some other point in time. At the core, we want to merge the SQL library into the main branch. For this, we need to unify the two separate SQL experiments that we currently have. Additionally, we need to clean up the experiments. For one of them, we are currently using Apache Calcite, which has both performance implications, as well as implementation implications. In my opinion we should remove the Calcite dependency, use some other parser and planner, such that we have more control over the evaluation. I am not sure if the current implementation actually works in a failure-free manner when executed distributed.
There has been some work on an SQL library which is experimental work. Currently, there are multiple versions and branches, see below for some references.
Context:
This is a very large issue, which needs to be broken down into smaller tasks at some other point in time. At the core, we want to merge the SQL library into the main branch. For this, we need to unify the two separate SQL experiments that we currently have. Additionally, we need to clean up the experiments. For one of them, we are currently using Apache Calcite, which has both performance implications, as well as implementation implications. In my opinion we should remove the Calcite dependency, use some other parser and planner, such that we have more control over the evaluation. I am not sure if the current implementation actually works in a failure-free manner when executed distributed.
See