Anybody out there having success with the NWorkspace pattern?

I've just begun to delve into my first experiments with Domain Drive Design and I'm taking advantage of the NWorkspace pattern. This pattern seems to make a lot of sense however I haven't been able to find very many examples of places this pattern has been successfully used or even publicly documented. Before I get to far into my implementation I would like to know if anybody has had success using this pattern or whether somebody could point me to any references where NWorkspace has been used in any open source project that I could learn from. Also are there better or more well known alternatives to this pattern that I should know about?

Brief background on NWorkspace

For those who may not be familiar with NWorkspace, it is a pattern introduced by Jimmy Nisson's which abstracts query and persistance responsibilities. In his book Applying Domain-Driven Design and Patterns, Jimmy Nilsson shows how NWorkspace can be use开发者_运维技巧d to abstract the infrastructure portions of a DDD Repository as well as provide a mechanism to perform cross Repository atomicity with regard to persistence.

It seems like he's recommending separate interfaces for read and write repositories.
I don't have experience with the pattern described, but I would recommend not having cross-repository transactions. Instead, I would suggest a few solutions popular amongst the DDD community (Eric Evans, Udi Dahan, Greg Young) that have really helped me:

1. Always use eager loading on your aggregate roots. Then you don't need cross-repository atomicity, and figuring out what's changed when you persist the object is a lot easier.
2. Use separate classes for writing (i.e. domain classes) and reading (i.e. your viewmodel). Create view model repositories that retrieve viewmodels directly from the database (instead of mapping domain objects to view model classes).

See if implementing the above 2 things simplifies your design.