SOFTWARE / SYSTEMS / AIEngineering news. Technical depth.
NEWS / Data · 2 MIN READ

MySQL 9.7.2 fixes deletion ordering in JSON duality-view updates

MySQL 9.7.2 corrects nested-row deletion during JSON duality-view updates and date conversion behavior, making targeted regression cases valuable.

Announcement: · From Oracle

MySQL’s July 28 release notes for version 9.7.2 describe a correction to JSON duality-view updates that delete nested child rows. The maintenance release changes operation ordering to avoid foreign-key violations in that path.

Nested document updates have relational consequences

The corrected implementation performs deletions before inserts and updates, processing those deletions in reverse binding order. The problem matters because a document-shaped update can affect several underlying relational rows with dependencies between them.

For teams adopting duality views, the useful regression case is a replacement document that removes a nested child while retaining its parent. Include deeper nesting and an intentionally invalid relationship, then inspect the underlying tables after both success and failure.

Our reading is that this patch deserves attention during feature evaluation as well as established production use. A single successful insert demonstrates little about how deletion, replacement and constraint handling behave together. Keep the expected relational state in the fixture rather than validating only the returned JSON.

Check conversions and existing data

The release also corrects a JSON_VALUE case involving a DATETIME return type, a default value and use in a DATE context. Another fix addresses incorrect values or skipped records when adding AUTO_INCREMENT to an existing table.

These are concrete reasons to test upgrade behavior with existing rows and edge-case inputs. For date expressions, compare predicate and join results when the default is selected, not just when the input contains a valid timestamp.

For schema changes, reconcile row counts and generated identifiers against a saved input dataset. The maintenance label should guide a focused verification plan: reproduce the application’s relevant boundary cases, compare outcomes, and retain those cases for subsequent upgrades.

SOURCES & CONTEXT

See the original announcement for availability and release details.