Re: import inconsistency

[Paul Edwards]

"Derek Robert Price" <derek@ximbiot.com> wrote in message 
news:mailman.7885.1055538119.21513.bug-cvs@gnu.org...
> >Should CVS simply be changed so that adding on branches is
> >treated the same way as imports, ie it creates BOTH the head
> >and the branch, both active?  This would have the effect of
> >fixing that inconsistency and moving it out of the Attic at the
> >same time.
>
> _NO_!  Adding to a branch does not create an active head.  A branch is
> separate from the head by definition.  Vendor branches are simply a
> special sort of branch that mean to act as a "base", for lack of a
> better word, of the head, unless a file has already been committed to
> the head by a user.

And what if my vendor branch is experimental?  E.g. USA sends
me a set of code to import to experiment with a particular
problem, only meant to be active for a couple of weeks.  Suddenly
it starts interfering with people working on the head, ie just doing
a "cvs checkout", no "-r".

Lucky for me I don't work on the head, eh?  :-)

> Thus the problem when a revision exists on the
> head, dead or not, requiring a merge.
>
> The case when there have been commits to the head before any import,
> when the import then conflicts, was never considered from the start and
> will need some planning and reworking of the code.

This is a separate issue.

Two issues:

1. conflict on "has been added".

2. imports creating active heads, branches not.

It sounds to me like all imports going to the Attic until requested
present on the head is the proper solution to (2).

BFN.  Paul.