They're used to store the data from OpenJPAStateManager.getImplData() and getImplData(int field).
Those impl data slots, in turn, are available for use by StoreManager implementations for random bits of per-field or per-instance state that doesn't fit elsewhere but is useful for the StoreManager implementation. This is similar to the 'intermediate data' concept, which you can see in the get/set intermediate calls in that class. OpenJPA uses intermediate data when it loads "extra" data from the database that can be consumed later to materialize something, and impl data for data that should be around for longer than just while the field is not loaded. For example, when OpenJPA loads an instance that has a -to-one relation that is not eager, but not explicitly lazy, it will generate SQL that selects the foreign key data for the relationship along with the other primary-table data. That way, if you do lazily load the data later, OpenJPA can go directly to cache, or at least bypass the join, when loading the related data. That info is stored in an intermediate, since either it or the relation itself can be loaded at any given time. Based no code analysis, it looks like OpenJPA only uses impl data for BLOBs currently.\ -Patrick On 5/14/07, Daniel Lee <[EMAIL PROTECTED]> wrote:
Hi, I'm trying to understand OpenJPA open source code but having difficulties in figuring out the use of _impl and _fieldImpl fields in PCDataImpl. Could someone help describe these 2 fields? Many thanks in advance. Tsunfang
-- Patrick Linskey 202 669 5907
