[MERGE] Allow changing --[no-]trees with reconfigure (bug 145033)

[Matthew D. Fuller]

On Mon, Jan 19, 2009 at 09:30:29AM +0200 I heard the voice of
Marius Kruger, and lo! it spake thus:
> 
> yes, when you run it from within a WT

Ah, in a repo.  OK.  I see it.


> So its not as dangerous as I thought, but I still think it is
> undesirable to have my existing working tree removed by this
> command.

Agreed.


This seems to be a result of _plan_changes() being an _incredibly_
obtuse function.  Apparently it has no way to say "leave X the hell
alone, dammit".  Additional fun can be seen if you make a checkout in
a repo, then reconfig --with-no-trees (or --with-trees if it's already
no-trees); it unbinds the checkout as well.  Holy hell.

It exists with other things too; note that 'bzr reconfig --checkout'
will make a WT in the branch at the same time; I'd expect it to just
do 'bind'.

Frankly, I'm tempted to just do the checks and set the vars manually
in set_repository_trees(), and skip over the whole morass of
_plan_changes() completely.  In fact, I think I'll do just that.


-- 
Matthew Fuller     (MF4839)   |  fullermd at over-yonder.net
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.