[hibernate-dev] Batch version data support in Dialect

[Steve Ebersole]

Dialect#defaultProperties is a broken concept IMO.  It was an attempt to
avoid an explosion of methods such as the one you propose.  I personally
agree that the specific methods are better.

However, all that said.. I do not believe what you say is accurate.  At
least not looking at the code.  In SessionFactoryOptionsStateStandardImpl
we build a Map of settings as follows:

1) apply Dialect#defaultProperties
2) apply user settings over top of that

So the user's setting should override the Dialect default.  That's how
settings should always work - something a user explicitly sets in their
config should always have precedence.  Are you seeing a case where that
does not happen as described?

Also, to clear up a misnomer... we don't "change the environment setting".
That's not how SessionFactoryOptionsStateStandardImpl and the config map
building works.

On Mon, Feb 8, 2016 at 11:58 PM Vlad Mihalcea <mihalcea.vlad at gmail.com>
wrote:

> Hi,
>
> While browsing the PRs on GitHub, I stumbled on this issue:
>
> https://hibernate.atlassian.net/browse/HHH-10290
>
> In the current implementation, the hibernate.jdbc.batch_versioned_data
> property is set to true and
> we override it at Dialect-level as follows:
>
> public Oracle12cDialect() {
>    super();
>    getDefaultProperties().setProperty( Environment.BATCH_VERSIONED_DATA,
> "true" );
> }
>
> Wouldn't it be better if the Dialect had a methods like:
>
> boolean supportsBatchVersionedData();
>
> and we wouldn't change the environment setting?
> With this method added, the user can override the Dialect setting using the
> environment variable.
> With the current implementation, we always override that setting.
>
> Vlad
> _______________________________________________
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/hibernate-dev
>