How to render SQL using NamingStrategies.toLower()

[dschulten]

NamingStrategies has a toLower() method to create a RenderNamingStrategy that applies toLower to table and column names, but I cannot find out how to activate it.

The RenderContextFactory has a setNamingStrategy Method where I could set NamingStrategies.toLower(), but I cannot find out how to tweak the RenderContextFactory - I tried to provide my own instance in an @Bean Configuration, but Spring Boot doesn't pick up my reconfigured RenderContextFactory.

Background: I use Postgres in Prod, but H2 during development, and both have different opinions about table and column naming.
Hence when running H2, I get this error:

Table "MANDANT" not found (candidates are: "mandant"); SQL statement:
SELECT "MANDANT"."ID" AS "ID", "MANDANT"."SCHLUESSEL" AS "SCHLUESSEL", "MANDANT"."BEZEICHNUNG" AS "BEZEICHNUNG" FROM "MANDANT" WHERE "MANDANT"."SCHLUESSEL" = ? [42103-232]

I managed to influence the RenderNamingStrategy by setting the dialect to JdbcPostgresDialect.INSTANCE in AbstractJdbcConfiguration, but that has unintended consequences elsewhere, e.g. when working with columns of Array type in H2.

[dschulten]

It turns out that the RenderNamingStrategy result is ultimately overridden by IdentifierProcessing. I found a way to tweak that by overriding Jdbc2H2Dialog.getIdentifierProcessing in my own dialect and force that tweaked dialect, but now my problem is that the dialect detection is effectively disabled - it will no longer use the JdbcPostgresDialect in production.

How can I make it so that spring boot detects the dialect automatically, but switches to lowercase table and column naming when using H2?

[schauder]

You really shouldn't use H2 for testing when you aren't using it for production. Use TestContainers instead.

Anyway: If you want to tweak the dialect detection there is a stack overflow answer how register your own custom dialect.