Skip to main content

[jsr338-experts] Re: JPA schema generation

  • From: Steve Ebersole <steve.ebersole@...>
  • To: jsr338-experts@...
  • Cc: Linda DeMichiel <linda.demichiel@...>
  • Subject: [jsr338-experts] Re: JPA schema generation
  • Date: Thu, 17 May 2012 08:56:39 -0500

My thoughts inline...

On 05/09/2012 01:36 PM, Linda DeMichiel wrote:

In conjunction with support for use of JPA in PaaS environments, we
will need to address the issue of schema generation. After we discuss
this area, I'd like to return to a broader discussion of multitenancy
support with this in view.


Here's what I think we'll need to address:

1. How/when schema generation happens:
a. at a deployment time phase (or predeployment time phase)
b. at EMF creation time

Well if the creation is driven by the platform provider, its 6-in-one IMO. But in standalone cases we obviously do not have the benefit of a platform provider, so there the answer has to be "at EMF creation".


2. Providing an option whereby the application can supply DDL for
schema generation as an alternative to the persistence provider
doing schema generation based on the ORM metadata

Is this per deployment?  Or per entity?

Another related thing to consider is "auxiliary database objects"; basically allowing create/drop definitions for database objects other than those explicitly named in mappings. Functions, procedures, view, etc, etc.


3. How, once schema generation has happened, tables can get loaded with
data. If we enable the application to provide DDL scripts for
schema generation, then supporting the use of SQL bulk insert scripts
would go hand-in-hand with this, but perhaps there are also other
approaches, e.g., bulk insert via JPQL(?).

+1  If you are exporting schemas, you will generally need this behavior.

I think SQL is much better than JPAQL here because of re-usability. I get the desire to insulate from object (table/column) naming, but still think the re-usability is more important.


4. The need for additional ORM metadata: e.g. metadata for index
generation; for foreign key constraints; other(?).

+1  I assume most vendors have this already in vendor-specific annotations.


5. More precise definition of existing metadata that can be used
for schema generation: what it comprises, and how it can/must
be used.

Not following this.  Could you elaborate?


6. Additional persistence.xml level metadata and additional metadata
the platform provider passes to the persistence provider to control
or configure the operations outlined above.

Again, not following.  What would some specifics be?


I am assuming for that Java EE 7 we need to support both the
separate database and shared database (aka separate schema) approaches.

Why not discriminator-based as well?


I am also assuming that it is the platform provider that does the
provisioning of either the database or the schema respectively (i.e.,
in accordance with whether the separate database or shared database
approach is taken). I.e., in the separate database approach, I'm
assuming that it is the persistence provider that creates the schemas
directly in the database or creates the DDL to generate the schemas
needed for the persistence unit. [I'll use the term "schema
generation" generically below to mean either table generation or
schema+table generation with this assumption in view.]


Here's a proposal to get the discussion started.

SCHEMA GENERATION

I think we should allow schema generation to be done either as a
deployment (or pre-deployment) phase or at createEMF time.

Deployment-time (or predeployment-time) schema generation could be
done either by the platform provider calling into the
PersistenceProvider implementation to generate the database schema or
by the platform provider running SQL DDL scripts.

SQL DDL scripts might be provided with the application (bundled as
part of the persistence unit) or the platform provider might invoke
the PersistenceProvider implementation to generate them from the
persistence unit metadata. If scripts are provided or generated, it
should be the platform provider's responsibility to run them against
the database. If the application has provided DDL scripts, the
application should also have the option to provide SQL scripts to do
a bulk insert of data. It would be the responsibility of the
platform provider to run these as well.

Personally, I think the ability to provide SQL init scripts for initial data insertion should be available regardless of whether the application provided the DDL or the persistence provider generated the DDL.


An alternative to the use of scripts is for the platform provider to
call into the PersistenceProvider to directly generate the tables in
the database. In the case that a database has been provisioned rather
than only a schema, the persistence provider would generate the
schemas and tables.

For the sake of argument, let's assume that we add a generateSchema()
method to the PersistenceProvider interface, which the platform
provider invokes during the deployment phase if SQL scripts have not
been provided by the application.

Information that needs to be passed in would include the following:

(1) Information that specifies how schema generation is to proceed.
This information could take the form of a SchemaGenerationInfo
interface or it could be in the form of additional standard properties.
(2) Connection to the database or schema. [Is this a Connection or a
DataSource ?]
(3) PersistenceUnitInfo


The SchemaGenerationInfo should include the following:

ddl-generation-mode: GENERATE_AND_EXECUTE | GENERATE_ONLY | EXECUTE_ONLY

GENERATE_AND_EXECUTE => generate ddl scripts and generate schema (if
applicable) and tables directly in the database

GENERATE_ONLY => generate ddl scripts

EXECUTE_ONLY => generate schema (if applicable) and tables directly in
the database

create-ddl-target: a Writer configured for the persistence provider
for outputting of the DDL scripts for the creation of schema/tables.
This should be null if EXECUTE_ONLY ddl-generation-mode is specified.

drop-ddl-target: a Writer configured for the persistence provider
for outputting of the DDL scripts for the dropping of schema/tables.
The should be null if EXECUTE_ONLY ddl-generation-mode is specified.

ddl-execution-mode: Used when executing directly against the database,
rather than generating scripts. Options should include:
CREATE_TABLES | DROP_TABLES | DROP_AND_CREATE_TABLES
Do we also need to distinguish CREATE_SCHEMAS_AND_TABLES ? DROP_SCHEMA ?
Or can these be implicit?
Do we also need an ALTER_TABLES capability here?

Schema evolution is infinitely more tricky than schema export and infinitely more tricky than it first appears. -1 to schema evolution capability IMO.


properties: These could include any vendor-specific properties


ADDITIONAL METADATA (persistence.xml)

If scripts are provided with the application (i.e., bundled as part
of the persistence unit), we should add entries in the persistence.xml
to identify them:
create-ddl-script
drop-ddl-script
data-load-script(s) [Do we need to support more than one load script?
If so, the ordering may need to be specified as well.]

Open issue: do we need metadata in the persistence.xml for
ddl-execution-mode as well? For example, this might be used in
standalone scenarios (?)


ORM METADATA (annotations and orm.xml)

(1) Indexes

I think we need to support indexes.

Here's a strawman annotation:

@Target({}) @Retention(RUNTIME)
public @interface Index {
String name() default "";
String columnList();
}

The columnList syntax could follow that of the OrderBy annotation:

columnList::= indexColumn [, indexColumn]*
indexColumn::= columnName [ASC | DESC]

If ASC or DESC is not specified, ASC is assumed

We'd also need to add
Index[] indexes() default {};
to Table, Secondary Table, CollectionTable, JoinTable, and TableGenerator


(2) Foreign Key Constraints

I see two possible approaches:

(a) Add a string-valued foreignKeyDefinition element to JoinColumn,
JoinColumns, MapKeyJoinColumn, etc. to specify a SQL fragment
(intended along the lines of columnDefinition) for defining a foreign
key constraint and/or for overriding the persistence provider's
default foreign key definition. It might also be helpful to allow a
name for the foreign key constraint to be specified for the case where
the provider is using its default foreign key generation strategy.

or

(b) Add a ForeignKey annotation to specify the foreign key constraint --
for example,

public @interface ForeignKey {
String name() default "";
ForeignKeyAction updateAction() default NO_ACTION;
ForeignKeyAction deleteAction() default NO_ACTION;
boolean deferred() default false;
}

public enum ForeignKeyAction {NO_ACTION, CASCADE, SET_NULL};

I have some misgivings about approach (b), and, given a lack of
database portability here, we might wind up needing a foreignKeyDefinition
fragment anyway.

Can you give an example of what you are thinking with 'foreignKeyDefinition'? Personally I was leaning towards (b) until I read this last bit. Do you just mean different capabilities (i.e., some dbs do not support deferred constraints, etc)?

There is an implication in terms of how a persistence provider needs to handle the case of cascading foreign keys in terms of mapped cascading. Take the case of remove operation, with ForeignKey.deleteAction==CASCADE all of a sudden the database has now physically deleted corresponding rows "underneath" the provider which has ramifications on the persistence provider in terms of cleaning up cache entries and possibly in terms of additional cascading. Yes, we have this mismatch today as is, but today we also don't have the needed visibility into that situation because we do not know the foreign key has been defined that way. Here that changes. Just want to make sure we then account for that in the other "entity operation"-related sections.

Either way, I think the ability to name the foreign key is definitely warranted.



(3) Other

What about the ability to distinguish a CHAR rather than VARCHAR
mapping for strings? Or should we just leave this as belonging in a
SQL fragment (with the understanding that we would be intentionally
discouraging the use of CHAR strings)?

(4) Anything else I'm missing?


APIs

I assumed the addition of a PersistenceProvider.generateSchema() method
above.

If we also support schema generation at EMF creation time, do we want
another createContainerEntityManagerFactory() method that takes a
SchemaGenerationInfo as well as PersistenceUnitInfo argument ?

If generation is done directly in the database, at some point the
container (or application) may need to have the schema/tables dropped.
I'm not sure of the best way to support this. A dropAndClose() method
on the EMF??





[jsr338-experts] Re: JPA schema generation

(continued)

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/15/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/15/2012

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/15/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/15/2012

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/15/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/21/2012

[jsr338-experts] Re: JPA schema generation

Rainer Kwesi Schweigkoffer 05/22/2012

[jsr338-experts] Re: JPA schema generation

Gordon Yorke 05/15/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/15/2012

[jsr338-experts] Re: JPA schema generation

Gordon Yorke 05/15/2012

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/17/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/18/2012

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/18/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/18/2012

[jsr338-experts] Re: JPA schema generation

Steve Ebersole 05/21/2012

[jsr338-experts] Re: JPA schema generation

Linda DeMichiel 05/22/2012
 
 
Close
loading
Please Confirm
Close