We can do it in 5.1.x branch.
I'll explain why this is important. When there is a high traffic on the site, then server admin may even need to restart MySQL server to solve the problem. In case, when ongoing agent is being executed at that time, then he doesn't react properly to that error and tries to run next sql with same error returned. This way, when error logging is enabled, then you'll get a lot (depending on sql count executed there) of errors there.
I've personally witnessed 2 millions of errors (agent was doing 1 query every second) in 2 days.