The SMRfs architecture and flow of control can understood at - SMRfs Architecture.
2. Code Structure:
The SMRfs code is pretty modularized. You can find details about the code structure at - Code Structure
3. How to build the code:
Building and running the SMRfs code is as simple as invoking a few scripts.
A detailed description of how to do this can be found at - Build and Run SMRfs
4. Regressions:
A good regression test would be one where we can ensure that -
Cleaning is not broken (that cleaning happens if required and does not corrupt data)
The best benchmark for checking regression would be the PostMark benchmark.
The instructions on running this benchmark and a sample results set can be found on
the Detailed SMR Benchmarks page. You can also try the other benchmarks for a more detailed analysis, if required.