Logging API

This file defines the desired API for an append-only durable log. A log like the one defined here is widely used in storage systems, such as a file system or a database, to make atomic crash-safe updates. In this lab, you will implement and prove correct the log itself, though the lab does not involve building anything on top of this log (like a file system). Logically, the state of the log is a list of blocks:

The log provides three operations: get, append, and reset. The get retrieves all of the blocks currently in the log:

append adds blocks to the log. The append is the most interesting part of the log, because append must be atomic with respect to crashes. That means that, if the system crashes in the middle of an append, either all of the blocks should be present (i.e., returned by get), or none of them should be present. append returns a boolean to indicate whether it succeeded or not. The intent is that append might fail if the log has grown too big on disk, and there's no room for the new blocks.

Finally, reset logically clears the log, removing all entries. This also must be crash-safe: if the system crashes in the middle of clearing the log, either the log contents should still be there, or the log should appear to be empty (according to get).