Naming Strategy

Naming Strategy

Jinx uses the Naming interface and the default implementation DefaultNaming to generate database object names.

What it names

The naming layer covers:

• join tables
• foreign-key columns
• primary keys
• foreign keys
• unique constraints
• indexes
• check constraints
• not-null constraints
• default constraints
• auto-generated constraints

Core rules

DefaultNaming uses a small set of predictable patterns:

• join table: jt_<a>__<b>
• foreign-key column: <owner>_<referenced_pk>
• foreign-key name: fk_<child_table>__<child_cols>__<parent_table>
• primary-key name: pk_<table>__<cols>
• unique name: uq_<table>__<cols>
• index name: ix_<table>__<cols>
• check name: ck_<table>__<cols> or ck_<table>__<constraint>
• not-null name: nn_<table>__<cols>
• default name: df_<table>__<cols>
• auto name: cn_<table>__<cols>

Normalization

Before building a name, Jinx normalizes each part:

• null becomes null
• non-alphanumeric characters become _
• repeated _ characters are collapsed
• names are lowercased
• empty results become x

Column ordering

For names that include multiple columns, Jinx:

1. normalizes each column name
2. sorts them case-insensitively
3. joins them with _

This keeps generated names stable regardless of declaration order.

Length limits

If a generated name is longer than the configured limit, DefaultNaming shortens it and appends a stable hash.

The result is:

<trimmed_prefix>_<hash>

The hash is based on SHA-256 and uses the first 8 hex characters.

Where configuration happens

JpaSqlGeneratorProcessor creates the naming setup during initialization.

It reads:

• jinx.naming.strategy
• jinx.naming.maxLength

The processor then stores:

• a JinxNamingStrategy for physical table-name transformation
• a DefaultNaming instance for generated database object names

So table names and constraint names are related, but they are not produced by the same API.