On Fri, 13 Jul 2001, Don Skanes wrote:
> There was a discussion a while ago about the infamous "evil twin"
> problem. Unfortunately I can't find the references to this so I
> apologize if I'm repeating the discussion.
>
> The "evil twin" phenomenom is where mulitiple elements exist for the
> same name and path which will happen if rather than performing a
> directory merge between branches to make an element visible, the user
> performs another mkelem for the element. What you end up with is an
> element (or should I say elements) with multiple histories ie. version
> trees.
>
> I would like to know if anyone has any tools to detect all occurences
> of this, and any tools to merge the twins together retaining the
> history.
I don't, but I could write one, but it would take some time... The idea is
that usually, evil twins will have a disjoint set of branches, so all that
needs to be done is to recreate the branches of one element in the
other. Nevertheless it is quite a task...
>
> I have a trigger enforced on mkelem at this point which disallows this
> but it is rather inefficient since it checks all directory versions of
> the parent of the new element for each new mkelem. I was wondering if
> anyone has found a better solution to this also.
It is usually good enough to simply check the LATEST on every branch and
the checkouts. This does open a window for maliciously adding and
unlinking a new element, "hiding" it from view, but I consider that far
fetched...
See also http://www.miaow.com/clearcase/faq/base_cc.html#1.2.8
-- cg- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
This archive was generated by hypermail 2b29 : Tue Jul 31 2001 - 22:03:54 EDT