karaf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benjamin Graf <benjamin.g...@gmx.net>
Subject Re: JDBC Feature
Date Sat, 18 Jan 2014 15:22:56 GMT
You right I missed the point that this is focused on xa only. There 
should maybe be opened a ticket at aries to also support non xa 
datasources as input. Not every jdbc driver has pooling capabilities itself.


On 18.01.2014 15:30, Jean-Baptiste Onofré wrote:
> This is different: aries.transaction.jdbc allows you to create a 
> "standard" datasource using a xadatasource.
> It doesn't provide any datasource (you have to provide at least a XA 
> Datasource).
> It's not the same usage.
> Where we can see an overlap it's in term of pooling as the Aries 
> RecoverableDataSource provide a pooling (using the Aries 
> ConnectionManagerFactory). That's true. But again, we can add a 
> template "Aries DataSource" that wrap a XADataSource to create a 
> pooled datasource. But you need the XADataSource ;)
> Regards
> JB
> On 01/18/2014 03:22 PM, Benjamin Graf wrote:
>> Hi JB,
>> well, I'm looking into which features/bundles gets installed how the
>> templates for specific providers does initialize the datasources (export
>> blueprint service) Take a look how they can be enhanced
>> (http://search.maven.org/#search|ga|1|aries.transaction.jdbc) by using
>> aries transaction jdbc.
>> Regards
>> Benjamin
>> On 18.01.2014 15:06, Jean-Baptiste Onofré wrote:
>>> Hi Benjamin,
>>> I'm not sure to follow you.
>>> Did you really take a look in the Karaf JDBC or JMS feature. It's
>>> mostly commands and MBean (not really any runtime providing). We
>>> completely leverage Aries Transaction for instance.
>>> It's the same for JNDI: we complete Aries JNDI (which doesn't provide
>>> any initial context) with a very simple XBean naming context.
>>> I'm not sure to see your point. Maybe you think that Karaf JDBC or JMS
>>> is more than that you think it is actually ;)
>>> Regards
>>> JB
>>> On 01/18/2014 03:02 PM, Benjamin Graf wrote:
>>>> Hi JB,
>>>> what I'm missing is a link to aries transaction jdbc which is in my
>>>> opinion the better template to provide out of the box. I think it's
>>>> worth thinking about a closer integration of this feature rather than
>>>> doing dbcp stuff for example. Well, of course that is a personally 
>>>> point
>>>> of view if you prefer to go with geronimo stuff providing jee APIs and
>>>> implementations or any other lib out there. :) May be it also worth
>>>> looking to integrate aries transaction jms for the karaf jms 
>>>> enterprise
>>>> feature. (Karaf vs. ServiceMix ?)
>>>> Regards
>>>> Benjamin
>>>> On 18.01.2014 10:52, Jean-Baptiste Onofré wrote:
>>>>> Hi Benjamin,
>>>>> if you take a deeper look, you will see that Karaf JDBC uses Aries
>>>>> (blueprint and transaction).
>>>>> Template files uses blueprint. The generic template uses DBCP, the
>>>>> other templates use specific database/JDBC driver capability. But 
>>>>> it's
>>>>> only template, you can deploy your own datasource definition.
>>>>> The Karaf JDBC module provides what is not in Aries: management of
>>>>> datasource, JDBC service, MBean and shell commands.
>>>>> So it's not different, it's a complement.
>>>>> Regards
>>>>> JB
>>>>> On 01/18/2014 09:40 AM, Benjamin Graf wrote:
>>>>>> Hi,
>>>>>> actually a lot of JDBC stuff has been done for karaf. I'm just
>>>>>> wondering
>>>>>> why it is done so heterogeneously in different apache projects. 
>>>>>> Karaf
>>>>>> uses blueprint and (not all the time!) dbcp for pooling and creating
>>>>>> datasources. Aries uses blueprint as well (:-)) but using
>>>>>> geronimo/tranql for pooling and creating. Since aries is part of

>>>>>> karaf
>>>>>> enterprise, why using yet another enterprise abstraction? By the
>>>>>> way. I
>>>>>> think aries has done a lot of good work to make working with JDBC
>>>>>> quite
>>>>>> handy.
>>>>>> Just my 0.02€
>>>>>> What do you think?
>>>>>> Regards
>>>>>> Benjamin

View raw message