Major Description
Here we track the work to add new golden file test support for the Spark analyzer. Each golden file can contain a list of SQL queries followed by the string representations of their analyzed logical plans.
This can be similar to Spark's existing `SQLQueryTestSuite` [1], but stopping after analysis and listing analyzed plans as the results instead of fully executing queries end-to-end. As another example, ZetaSQL has analyzer-based golden file testing like this as well [2].
This way, any changes to analysis will show up as test diffs, which are easy to spot in review and also easy to update automatically. This could help the community together maintain the qualify of Apache Spark's query analysis.
[1] https://github.com/apache/spark/blob/master/sql/core/src/test/scala/org/apache/spark/sql/SQLQueryTestSuite.scala
[2] https://github.com/google/zetasql/blob/master/zetasql/analyzer/testdata/limit.test.
Attachments
Issue Links
1.
|
Create analyzer golden file based test suite |
|
Open | Unassigned |
2.
|
Enable new golden file test framework for analysis for all input files |
|
Resolved | Daniel |