[SPARK-35560] Remove redundant subexpression evaluation in nested subexpressions

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.2.0
• Fix Version/s: 3.2.0
• Component/s: SQL
• Assignee: L. C. Hsieh
• Reporter: L. C. Hsieh
• Created: 28/May/21 23:11
• Updated: 02/Jun/21 02:13
• Resolved: 02/Jun/21 02:13

Description

In some cases, whole-stage codegen's subexpression elimination will do redundant subexpression evaluation. We should reduce it. For example, if we have two sub-exprs:

1. simpleUDF($"id")
2. functions.length(simpleUDF($"id"))

We should only evaluate `simpleUDF($"id")` once, i.e.

subExpr1 = simpleUDF($"id");
subExpr2 = functions.length(subExpr1);

Issue Links

• links to: [Github] Pull Request #32699 (viirya)