In Greg's devTeach presentation he brings up the concept of a write-ahead log, which is a powerful, yet simple mechanism for temporary event storage which is also a capability offered by virtually all relational databases.
In a database scenario, the concept is simple: whenever a command is performed, it is performed in memory and immediately recorded in the log. Then, if the database process accidentally restarts/shuts down, etc. it can be brought back to a known state.
This is where the comparison ends. In a database world, uncommitted transactions are added to the log. In DDDD, only committed events are added to the log as part of the transaction. This is a very important difference.
In DDDD, when the system comes back online after a restart, we want it to be as if none of the uncommitted work ever happened. Therefore, only committed events are added to the write-ahead log.
The reason I'm posting on this is that at about 40:10 in the presentation he makes the statement "this one is actually committed" which makes it appear as though uncommitted events could be added to the write-ahead log. Earlier in the presentation--at 39:40--he says, "I am taking my transactions as they're getting committed and just putting them into this log."