Naming strategies

Hibernate turns entity and field names into table and column names, and each framework configures that differently. Migrax must use exactly the same rules, or the generated tables won't be the ones your application looks for.

Strategy	Used by	homepageURL	fullName	Join table for Product.tags
spring	Spring Boot	homepageurl	full_name	product_tags
jpa	Quarkus, Helidon, Jakarta EE, plain Hibernate, Micronaut without Micronaut Data	homepageURL	fullName	Product_Tag
micronaut	Micronaut Data	homepage_url	full_name	product_tag
jpa-snake	Hibernate with only a snake_case physical strategy	homepageurl	full_name	product_tag

Names you give explicitly with @Table(name = ...) or @Column(name = ...) pass through the same physical rule: Spring Boot and Micronaut Data turn them into snake_case too.

How Migrax chooses

1. An explicit setting: --naming, MIGRAX_NAMING or migrax.naming.
2. The naming recorded in .migrax/snapshot.json by the first generate. This keeps names stable even if the configuration changes later, for example when the database URL moves into environment variables.
3. A naming strategy configured for Hibernate, such as spring.jpa.hibernate.naming.physical-strategy or hibernate.physical_naming_strategy.
4. The framework's default, detected from the build file and configuration.

migrax init and migrax doctor show the result. When the snapshot's naming differs from what your framework uses, doctor warns and explains how to switch.

Changing the naming of an existing project

Switching naming renames every affected column. Set migrax.naming, run migrax generate, and accept the renames it proposes, so the data is kept.

Sequences

Generated ids (@GeneratedValue with AUTO or SEQUENCE and no named generator) use:

Hibernate	Sequence
6 and 7	<entity>_SEQ per entity, allocation size 50
6 and 7 with Micronaut Data	one hibernate_sequence
5.x	one hibernate_sequence, allocation size 1

On MySQL, which has no sequences, Hibernate uses a one-row next_val table of the same name, and so does Migrax.