On Feb 25, 2010, at 11:24 AM, Andreas Schwab wrote:
>> Would the equivalent of "git reset olderrevid ;  git push -f" inconvenience people very much?
> 
> Don't even _think_ about that!

I hear you.
The consequences get worse in proportion to the time span between the faulty commit and the history manipulation.

What happens if somebody checked in code for which permission hasn't been obtained and can't be obtained?  Suppose the copyright-holder requests removal.