Specialised SQL Backend

[Robert Staudinger <ro...@salzburgresearch.at>]

Hello,

while evaluating frameworks for the creation of an image repository i 
came across jackrabbit and the standards it implements. Thing is i have 
to geographically index (e.g. with PostGIS) the objects in my database, 
the best approach however is not yet clear to me.

As far as I gathered from the documentation it's not recommended to 
interpret the node content in a PersistenceManager implementation, is 
there any other way (higher level interface maybe) that could be used 
for passing such "hints" down to the backend?

Or could you rather recommend that frontend and backend know how to 
handle certain node types/attributes while jackrabbit in the middle doesn't?

Thanks for any hint.

Best regards,
Rob

Re: Specialised SQL Backend

[Robert Staudinger <ro...@salzburgresearch.at>]

Stefan Guggisberg wrote:
> hi rob,
> 
> On 4/13/06, Robert Staudinger <ro...@salzburgresearch.at> wrote:
>> Marcel Reutegger wrote:
>>> Hi Rob,
>>>
>>> I'm not sure if that is of any help, but you could use an EventListener
>>> to get notifications about changes in the workspace and then trigger
>>> your indexing process.
>> Oh, there seems to be a misunderstanding. I want to use an SQL backend,
>>   RDBMS usually don't need any active indexing triggering/management
>> from outside on their tables.
>> Having slept over my original post i think the question is more like
>> "Would it be possible to hack an SQL backend that allows for the storage
>> being optimized towards the application by using indices and
>> distributing tables in a favourable way".
> 
> sure, you could *hack* your own application specific db persistence manager
> for your purposes. but you have to be aware that this approach is not
> officially supported by jackrabbit which is designed to support arbitrary
> node types/structures. future changes in jackrabbit could break your code.

Ok.
Continuing my search for how to extend jackrabbit in a supported and 
forward compatible way I came across the node type registry interface. 
Is this the recommended way for enforcing application specific schemas? 
Having custom nodes also seems like a clean way to me letting an app 
specific PM know about preferred storage and indexing requirements.
Or am I completely off track?

Best,
Rob

Re: Specialised SQL Backend

[Stefan Guggisberg <st...@gmail.com>]

hi rob,

On 4/13/06, Robert Staudinger <ro...@salzburgresearch.at> wrote:
> Marcel Reutegger wrote:
> > Hi Rob,
> >
> > I'm not sure if that is of any help, but you could use an EventListener
> > to get notifications about changes in the workspace and then trigger
> > your indexing process.
>
> Oh, there seems to be a misunderstanding. I want to use an SQL backend,
>   RDBMS usually don't need any active indexing triggering/management
> from outside on their tables.
> Having slept over my original post i think the question is more like
> "Would it be possible to hack an SQL backend that allows for the storage
> being optimized towards the application by using indices and
> distributing tables in a favourable way".

sure, you could *hack* your own application specific db persistence manager
for your purposes. but you have to be aware that this approach is not
officially supported by jackrabbit which is designed to support arbitrary
node types/structures. future changes in jackrabbit could break your code.

cheers
stefan

>
> Best,
> Rob
>

Re: Specialised SQL Backend

[Robert Staudinger <ro...@salzburgresearch.at>]

Marcel Reutegger wrote:
> Hi Rob,
> 
> I'm not sure if that is of any help, but you could use an EventListener 
> to get notifications about changes in the workspace and then trigger 
> your indexing process.

Oh, there seems to be a misunderstanding. I want to use an SQL backend, 
  RDBMS usually don't need any active indexing triggering/management 
from outside on their tables.
Having slept over my original post i think the question is more like 
"Would it be possible to hack an SQL backend that allows for the storage 
being optimized towards the application by using indices and 
distributing tables in a favourable way".

Best,
Rob

Re: Specialised SQL Backend

[Marcel Reutegger <ma...@gmx.net>]

Hi Rob,

I'm not sure if that is of any help, but you could use an EventListener 
to get notifications about changes in the workspace and then trigger 
your indexing process.

regards
  marcel

Robert Staudinger wrote:
> Hello,
> 
> while evaluating frameworks for the creation of an image repository i 
> came across jackrabbit and the standards it implements. Thing is i have 
> to geographically index (e.g. with PostGIS) the objects in my database, 
> the best approach however is not yet clear to me.
> 
> As far as I gathered from the documentation it's not recommended to 
> interpret the node content in a PersistenceManager implementation, is 
> there any other way (higher level interface maybe) that could be used 
> for passing such "hints" down to the backend?
> 
> Or could you rather recommend that frontend and backend know how to 
> handle certain node types/attributes while jackrabbit in the middle 
> doesn't?
> 
> Thanks for any hint.
> 
> Best regards,
> Rob
> 
>