Bad merge base selection

[Martin Pool]

On 17/09/05, Aaron Bentley <aaron.bentley at utoronto.ca> wrote:
> This is happening because the complete bzr tree has two root nodes;
> thre is a revision committed by Martin whose revision-id ends with eeae
> (sorry to be vague, I have to boot windows to use my modem).
> 
> This revision has no parents, which makes it a second root node.
> My code assumes there's only one root node.
> I'll fix it by using the null revision as the root of every graph,
> unless someone has a better idea.

That sounds like a good idea.

-- 
Martin