Thats one of the things I always liked about MySQL. The logs contained essentially just plain old SQL. If you needed to undo something that someone did you could just restore/recover and skip that statement in the logs. Unless subsequent changes were affected by that change (which was rare) you could back what you needed pretty easily.
I'm pretty beat right now from a rough week, but I can't see how that kind of logging can ensure rigorous data recoverability in the event of a restore. Different hardware can have VASTLY different performance characteristics (or even the same hardware depending on concurrent usage) and just allowing a stream of sql statements to replay sure sounds like it could allow stuff to COMPLETE out of order, leading to invalid data states.
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service