rollback revisited

[Derick H Siddoway]

Hi.

I've seen rollbacks discussed here, so I hate to rehash old stuff,
but I'm currently rereading the manual, and a question occurred to me.  
In all of the discussion, both here and in the manual, regarding 
transactions, the emphasis seems to be on preventing interference 
from other threads.  In this light, using a LOCK TABLES / UNLOCK 
TABLES sequence certainly makes sense.

However, none of the db developers I work with use it like that.
They tend to be doing something like writing a script to do a
mass UPDATE, and then, when it doesn't work as expected, they
issue a ROLLBACK and try again.

How do you emulate this with mysql?

Thanks -

derick