aries-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Giuseppe Gerla <giuseppe.ge...@gmail.com>
Subject Re: About ARIES-1023 and ARIES-1079
Date Sun, 21 Dec 2014 09:24:44 GMT
Hi JB
i'm still working to find a better solution for this point...
The problem is that PersistenceBundleManager create instance of
EntityManagerFactory parsing persistence.xml. So when we process the
blueprint xml file it's too late.

I think that this is not a correct approach. We are in a managed container
so from my point of view the EntityManagerFactory must be instantiate
during the parsing of blueprint xml file. If we have a persistence unit
declared in persistence.xml and not in blueprint, the container will not
create the entitymanagerfactory.

Wdyt?

Anyway implement my proposal is quite complex ... I'm working on that.


Regards
Giuseppe


2014-12-17 16:37 GMT+01:00 Jean-Baptiste Onofré <jb@nanthrax.net>:

> Hi Giuseppe,
>
> it sounds good to me. Let me give some time to review your changes deeper.
>
> Regards
> JB
>
>
> On 12/09/2014 01:53 PM, Giuseppe Gerla wrote:
>
>> Hi
>> I'm working on two issues in object. I linked them because they are
>> duplicated.
>> The problem is a small bug in the parser that I resolved to commit
>> 56bf3dffd15be41e1e7151993d495c4ef19a87c1 on my fork
>> https://github.com/ggerla/aries/.
>> Please merge on official trunk and close the issue.
>>
>> I believe that you will have to open another issue because properties that
>> are passed to JTAEntityManagerHandler are NOT used to create the
>> EnetityManager.
>> I try to explain better... Properties specified in jpa:map section are
>> passed to JTAEntityManagerHandler to create an EnetityManager "managed"
>> with the call
>>
>> manager = emf.createEntityManager (props);
>>
>> where emf is an instance of QuiesceEMFHandler. The EntityManagerFactory
>> implementation first create an EntityManager and then add properties
>> passed. So properties are NOT used for the creation.
>> I created a test to show that this does not work.
>> I also try to fix this problem with
>> commit 80bcaca80a09dd088ba25151376b68b646cffb0e. I do not like it so much
>> but it could be a start point.
>>
>>
>> Wdyt?
>>
>>
>> Giuseppe
>>
>>
> --
> Jean-Baptiste Onofré
> jbonofre@apache.org
> http://blog.nanthrax.net
> Talend - http://www.talend.com
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message