Hi all,

I'm trying to understand how exactly works the nonstrict read-write strategy of the second level cache. Unfortunately, I can't find any details about this implementation, all documentation found on the web says: "Needed if the application needs to update data rarely..." but does not really explain how and when the cache is updated.
There is an excellent article about the read-write strategy http://clustermania.blogspot.com/2009/0 ... level.html that explains the details of the implementation. If any one could provide any link or info for an equivalent about the nonstrict read-write strategy that would be really great!
Basically what I'm wondering about nonstrict r/w strategy is:
- If the update of a cached object is done regardless of a commit and transaction/cache entry time stamp (if any used for this strategy). For example Transaction 1 is reading object A, transaction 2 update Object A but has not commit yet then transaction 1 read again object A: What data does it get ? The updated object A by transaction 2 (not yet committed though) ?
- What rules are used for maintaining cached object state in a concurrent access environment ?
I'm wondering why there is no detail explanation about nonstrict read-write strategy and how people (you guys) know how to use it then ? Or maybe I've just missed the right paper about it or it is just more simple than I think...

Sorry for the long post and thanks in advance for any info/link on the subject...

Arnaud.