WIP Built some initial e2e datafusion infrastructure #10
Conversation
There was a problem hiding this comment.
Should we call this optd-datafusion instead or something more descriptive than "infra"?
Make sure to fix all of the clippy lints (you can either look at the Files changed tab on GitHub or just run cargo clippy --all-targets
Also, please add comments on everything because I can't really understand the code...
See the official Rust style guide as well for things that clippy cant check (things like full punctuation on comments)
|
It might be a good idea to fully rebase everything on top of main, something like |
|
Yes, I agree. Most of what was done here has become useless. However, I think I'll still try and make a version of this stuff work on this branch and then worry about moving stuff around later. |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files
|
we don't want to support it for now
|
Integrated into #26 |
## Problem With the initial representation and storage added in #4 and #22, we now want to support the full pipeline going from parsing SQL, optimizing the plan using optd, and executing the query in Datafusion. ## Summary of changes - Integrate all @SarveshOO7's good work in #10 - Added one mock physical implementation rule + operator for each logical operator - Refactor scalar operator storage and reduce code bloat. - Add physical storage tables and memo API. - Bump MSRV to 1.81.0 to be compatible with datafusion 45.0.0: apache/datafusion#14330 --------- Signed-off-by: Yuchen Liang <[email protected]> Co-authored-by: SarveshOO7 <[email protected]>
Problem
We want to be able to run our queries on datafusion.
Summary of changes
The only update I have is that I have setup the datafusion e2e to be able to run multiple queries and print the results and the plans. Currently, since we do not have an optimizer built out and we don’t have an logical plan IR to convert the datafusion logical plan into, I am simply running the datafusion optimizer.