Ability to Control Schema and Factory Class Names about jooq

[lukaseder]

I'm not sure if I understand. Today, there is the SchemaMapping functionality as documented here:

http://www.jooq.org/manual/ADVANCED/SchemaMapping/
http://lukaseder.wordpress.com/2011/11/09/development-schema-production-schema/

Besides, there is an open feature request to give users of jooq-codegen more flexibility in choosing their own naming schemes:

https://sourceforge.net/apps/trac/jooq/ticket/618

But a much more interesting approach was proposed by a user who is currently implementing a Groovy Templating extension. I'm also thinking about an XSLT code generator that transforms an intermediate XML schema definition:

https://sourceforge.net/apps/trac/jooq/ticket/911

Are any of those ideas what you had in mind? In any case, thanks a lot for the feedback! Feel free, to participate also in the news group to get discussions with other users going:

https://groups.google.com/forum/#!forum/jooq-user

Cheers
Lukas

[dacc]

Hey Lukas,

Thanks for your reply.

We use system properties to specify where the database we want to generate code for is. I guess I'm looking for an option to explicitly specify the class name for the schema in the generated code so that this can always be reliable, regardless of what database is being used. Something like schemaClassName here:

...
                    <jdbc>
                        <schema>${schemaLocalName}</schema>
                    </jdbc>
                    <generator>
                        <target>
                            <schemaClassName>SchemaReliableName</schemaClassName>
                        </target>
                    </generator>
...

This solves the compile time half of the problem that mapping solves for us at runtime.

Does that explain it better?

Thanks,

• Dan

[lukaseder]

I see, that's an interesting idea. I filed it as feature request #958:
https://sourceforge.net/apps/trac/jooq/ticket/958

Can you elaborate a bit more on your use-case, so I can get this right, especially in the documentation. In runtime SchemaMapping, it's possible to re-write schema names and table names. The use case is usually that the development or integration test schema (= code generation schema) has different namings from the (various) production schemata.

When does it happen that the code generation schema is different? Do you build the generated classes several times, once in every environment? Do you think, re-writing table names at code generation time is useful, too?

[dacc]

For us the schema / database is a matter of environment configuration. Each developer may have several databases with the same DDL loaded, but different data. Also our integration server runs many builds that require databases for their tests, and we may have e.g. two different branches of the same app with their own build jobs. In this case they'll need their own databases too, even though the table structure is the same.

[lukaseder]

I see, that makes sense. In the short term, I can see how your suggested feature will help. It should be easy to implement in jooq-codegen

In the long term, though, I prefer what was discussed in this thread here with Sergey Epik:
https://groups.google.com/forum/#!topic/jooq-user/4-iZMVDtkGA

This is about separating the code-generation phase from the schema introspection phase, meaning that the schema meta-data is first transformed into XML (or some other format), whereas code generation is based precisely on that database-independent XML source.

[lukaseder]

I'm closing this GitHub issue. It will be further tracked in Trac:
https://sourceforge.net/apps/trac/jooq/ticket/958

[lukaseder]

The configuration will be

                        <configuration>
                            <jdbc>
                                ... the schema here is deprecated. it'll be removed later
                                <schema>${schemaLocalName}</schema>
                            </jdbc>
                            <generator>
                                <database>
                                    ... the input schema is used for reading data in jooq-meta
                                    <inputSchema>${schemaLocalName}</inputSchema>
                                    ... the output schema is used for code generation
                                    <outputSchema>SchemaReliableName</outputSchema>

I'll create a sample configuration in the jooq-codegen-maven-example Maven artefact. Both code generation and runtime will be integration tested in a new integration test suite.

[dacc]

Awesome, looks great.

Thanks!