Open alamb opened 2 weeks ago
There are a few tests that don't pass for reason's I haven't debugged yet but it looks like recomputing schemas may be masking issues elsewhere. Once https://github.com/apache/datafusion/pull/10356 is merged I'll do another profiling run to figure out if recomputing schema shows up in the traces and will keep hacking on this if so
Update here is I don't think I'll have a chance to work on this in the near term unless recomputing the schema shows up in planning benchmarks again.
I think given the level of effort involved, this approach is not likely to yield a large speedup (though I do think this PR's approach would be an improvement over the current code)
Draft as it builds on https://github.com/apache/datafusion/pull/10356
Which issue does this PR close?
closes https://github.com/apache/datafusion/issues/10365
Rationale for this change
recomputing the Schema can be expensive, so only do it when necessary
What changes are included in this PR?
Transformed::yes
)This is a big change change as it needs to is properly track when expressions were changed which required updating the entire expression rewrite
Are these changes tested?
Existing CI
Are there any user-facing changes?
No functional changes
TODO performance tests