[hibernate-dev] [OGM] Thoughts for the Infinispan / Hot Rod dialect

[Sanne Grinovero]

On 29 July 2015 at 10:35, Davide D'Alto <davide at hibernate.org> wrote:
>> I think that in the case of Hot Rod clients we should not use
>> AtomicMaps
>
> Out of curiosity, what about the grouping API?
> We asked the Infinispan team to add some methods to it because we might use
> them
> to implement the dialect over HotRod. These methods are still missing in
> the HotRod protocol, though;
> as far as I know.

Right, essentially my proposal is to avoid the grouping API.
 - It's not implemented yet
 - Doesn't give us the same locking semantics which we liked for
embedded mode (no transactions)
 - If it can query we don't strictly need it anymore

>
>> Would you agree on try basing the approach on a brand new dialect
>
> I'm fine having a separate dialect.
>
>> and on protobuf schema generation?
>
> What kind of information do you need to generate the schema?

All attributes and types, and types of relations. Shouldn't be very
different than a RDBMs schema.

> Is it something that you can do using the information we provide via the
> interface SchemaDefiner (and BaseSchemaDefiner) at startup?

I expect so. But when we define the Protobuf templates, these are not
just applied to Infinispan but the persisters should also hold on
them, as they will need to use the same protobuf encoded messages on
each CRUD.

Thanks,
Sanne