[Monotone-devel] Re: read/write permissions

[Bruce Stephens]

Boris <address@hidden> writes:

[...]

> Out of curiosity and to understand monotone better: Do you see then
> two  heads in branch B?

Yes.

> Or do heads need to have the same ancestor cert somewhere?

No.

Not currently, anyway.  Identifying (at a low-level) branches by their
names is (like identifying keys by their names) probably a mistake,
and probably it'll be resolved as part of the "policy branch" stuff.
I believe that's the plan, anyway.  I don't know much more than that,
but perhaps it'll involve all revisions in a branch having a common
ancestor.

Having an odd branch cert in the wrong place doesn't necessarily cause
any immediate problems: "mtn update" will still work, for example,
because it only looks at descendents of the workspace base revision.
But it would screw propagate and merge.