[hibernate-dev] SingleTableEntityPersister memory footprint in 5.3

[Andrej Golovnin]

Hi Sanne,

> you're right, I found the most interesting email from you now (reading
> the whole thread) :
>  - http://lists.jboss.org/pipermail/hibernate-dev/2012-May/008410.html
>
> I'm not sure why the discussion stopped, I'm sorry.
> Let's try to fix it this time, to make sure of that we need a JIRA.
> Someone created one? Guillaume?
>
> summary:
>  - problem introduced by https://hibernate.atlassian.net/browse/HHH-4546
>  - great results applying Andrej's
> https://github.com/golovnin/hibernate-orm/commit/1cc8a23d1d5f4079fb090cca63a44e8660f05c7f
>  - the patch could be improved today with Java8
>    -- I'd like to have some degree of upfront validation at boostrap still
>    -- hopefully explore an array based structure
>    -- do not extend CHM
>    -- ideally have no read barriers at all at runtime
>

This is an old patch. The current version does not extend CHM and even
does not use it. It adds ca. 30 lines of code and modifies two classes
to achieve the most memory savings.

And Sanne, it is not needed to be sorry. Thanks to GIT I can apply my
patches to the new version of Hibernate without big problems and have
the best Hibernate version ever. :-)

Best regards,
Andrej Golovnin