Ben Wing <ben(a)666.com> writes:
I'm to blame for the multiple-change stuff.
No blame is due. In this case, I didn't mean to imply that the code
was unreadable. The code is good, it's just that every complexity
takes time to understand.
Someone might want to verify that it actually improves efficiency.
True enough. But then again, I really wish the same kind of
optimization can be employed for undo. M-x query-replace is
*horribly* slow with large files (even on very fast machines), and I
think much of that slowness comes from repetitive small undo records.