On Fri, Aug 22, 2014 at 12:26:12PM +0200, Bert Huijben wrote: > > I'm not sure this should even be flagged as a conflict. > > However, the victim node kind is wrong. > > I think it should say local file delete, ... > (but local, left-src and right-src are all victims) As already mentioned on IRC, when we introduced the term "tree conflict victim" in 1.6 it always referred to the local node. It never referred to any incoming ones. I think we should keep this distinction to avoid misunderstandings. The left-src and right-src cannot be victims. Whatever the user saw at the affected path in their working tree, before running the operation, is the victim.
> On a local delete we should really look one layer below the current op-depth > to see what the original node kind is/was, before the local > delete/replacement. I think we should record the victim's node kind when discovering the conflict, so we won't have to guess it later.
