Re: [eclipselink-users] Re: rcp, osgi, and dynamic weaving

["Ted Pricer" <ted.pricer@xxxxxxxxx>]

I can try to test it with 3.4, although the HookRegistry is tagged as
since 3.2 (and I have access to that API, so it clearly was in 3.3).
I'll let you know the results from updating to 3.4.

Ted

On Tue, Jan 13, 2009 at 2:47 PM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:
> Hi Ted,
>
>  I believe the hook APIs were added in Equinox in Ganymede (3.4) - though
> someone can correct me if I am wrong.
>
>  We certainly have not tested the dynamic weaving on a pre-Ganymede version.
> The best bet for users on earlier versions is to use static weaving.  Here
> is some info about our weaving support:
>
> http://wiki.eclipse.org/Using_EclipseLink_JPA_Extensions_%28ELUG%29#Using_EclipseLink_JPA_Weaving
>
> -Tom
>
>
>
> Ted Pricer wrote:
>>
>> Tom,
>> Maybe. Did you mean:
>> org.eclipse.persistence.jpa.equinox
>> org.eclipse.persistence.jpa.equinox.weaving
>> which were both referenced in this thread.
>>
>> Assuming that is what you meant, then I can provide a few more details
>> that may be important:
>> I've modified the org.eclipse.persistence.jpa.equinox.weaving manifest
>> file to use bundle-version 3.3.0 (my current target platform).
>> I've set breakpoints in the WeavingConfigurator class, which are not hit.
>>
>> Are you aware of anything that could be preventing the discovery of
>> the HookConfigurator?
>>
>> Ted
>>
>> On Tue, Jan 13, 2009 at 1:24 PM, Tom Ware <tom.ware@xxxxxxxxxx> wrote:
>>>
>>> Appologies... answers to my emails in the wrong order.
>>>
>>> What I meant to say for this email is that I do expect this to work.
>>>
>>> There should be two bundles that are deployed to allow this to work.
>>>
>>> org.eclipse.persistence.jpa.weaving is a component of EclipseLink and
>>> org.eclipse.persistence.jpa.weaving.equinox is a fragment for Equinox.
>>>
>>> Are you using both of these?
>>>
>>> -Tom
>>>
>>> Tom Ware wrote:
>>>>
>>>> Hi Mike,
>>>>
>>>>  One question, one comment:
>>>>
>>>> - In EMImpl.flush(), you do a check to see whether we are within a flush
>>>> before toggling the withinFlush variable, but the call to write is not in
>>>> the same if statement.  Are there cases where we could get into flush() with
>>>> withinFlush set to true where we would still want writeChanges to execute?
>>>> - In EJBQueryImpl.executeReadQuery, you directly access the withinFlush
>>>> variable in entityManager.  Stylistically, I would prefer to see that access
>>>> through an accessor method.
>>>>
>>>> -Tom
>>>>
>>>> Ted Pricer wrote:
>>>>>
>>>>> Does this currently work with an RCP based application? I've tried
>>>>> including the fragments in my project but haven't seen any dynamic
>>>>> weaving taking place. In fact the WeavingConfigurator isn't being
>>>>> instantiated (it looks to me like that is where the weaving libraries
>>>>> get incorporated at runtime).
>>>>>
>>>>> I'm currently launching via an eclipse debug configuration.
>>>>> _______________________________________________
>>>>> eclipselink-users mailing list
>>>>> eclipselink-users@xxxxxxxxxxx
>>>>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>>
>>> _______________________________________________
>>> eclipselink-users mailing list
>>> eclipselink-users@xxxxxxxxxxx
>>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>
>> _______________________________________________
>> eclipselink-users mailing list
>> eclipselink-users@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>