## Mapping strategy
I think we have explored three main options while implementing the association mapping in
MongoDB
1. Put the assoc info within the entity document we navigate from
2. Put the assoc info in a dedicated document and dedicated collection
3. Put the assoc info in n documents (one doc per row) and a dedicated collection
And slightly orthogonal to this, we have considered to use a prefix for the association
collection to make sure it does not clash when mapping @ManyToOne @JoinColumn.
I am reading MongoDB in Action and it seems option 1 is the most natural for MongoDB. At
least let's offer the option via a property collected by the datastore / dialect. We
can decide to reduce options down the road we we have a better grasp. And that will
prevent rewriting the dialect every tie someone changes his mind (like me ;P)
## Optimizing what is store
We definitely store the same data over and over (like the table in association rows. At
some point we probably will want to clean that up to do a more "natural"
mapping.
(More) thoughts on associations?
Emmanuel
PS: this is nothing new but I'd rather capture this info by mail rather than via IRC
discussions to consolidate it.
Show replies by date