This would be an interesting project. That is, take the block storage interface in PyORAM and create a wrapper class that provides a filesystem interface (perhaps a backend for PyFilesystem). One would need to:
- implement reasonable caching strategies for frequently used blocks
- keep track of unused blocks
- efficient mapping of file objects to blocks (perhaps with block sub-allocation strategies)
- etc.
This is somewhat duplicating work that would be handled automatically if we were implementing a kernel driver, but the point of this project is to avoid development in C.
This would be an interesting project. That is, take the block storage interface in PyORAM and create a wrapper class that provides a filesystem interface (perhaps a backend for PyFilesystem). One would need to:
This is somewhat duplicating work that would be handled automatically if we were implementing a kernel driver, but the point of this project is to avoid development in C.