Re: [RFH] revision limiting sometimes ignored

On Mon, 4 Feb 2008, Linus Torvalds wrote:

Not really tested at all, not really thought through. And that recursion 
avoidance could be smarter.

.. and by "could be smarter", I obviously mean "really *really* should be 
smarter". Because this is O(2**n) in the number of merges, which is not 
acceptable even if the constant is really small.

So I really don't mean that we should do it this way. The right thing to 
do would be to add a new object flag for that "connected to UNINTERESTING" 
property, and setting it as we traverse the graph in that 
"commit_is_connected()" logic. That should get rid of the exponential 
behaviour.

			Linus