On Fri, 8 Aug 2003, Daniel Vainsencher wrote:

> A couple of questions here -
> * I don't know if that is the intent, but it would seem quite natural
> that the repository list is the place where I manage all my repository.
> If so, why should I be unable to use it until I've defined packages? I
> should be able to add a repository, and then open a version from it.

Yes, you should.  This is basically a bug (I could go into the details of
why it works the way it does but since I"m not interested in defending
them I won't).

> * In the same vein, a Merge button would be useful there.

Yes, this might be useful, although there is part of me that prefers the
two step open, then merge - it's only one extra click, and it reinforces
the object model somewhat (merge is an operation on a Version, *not* on a
Repository).  I'm on a bit of a Naked Objects kick with the Monticello UI
(expect to see things further in that direction later).

> BTW, the
> innocent seeming label "load" is IMHO not scary enough for the operation
> it stands for IIUC - "lose all my changes, replacing them with the
> loaded package contents". Best I can find is "revert to...", but it
> isn't very precise.

I think I agree with Julian here - the best thing is simply to warn people
before proceeding.

> * I fixed some code before loading monticello, then loaded monticello.
> Then I merged a previous existing MCV. In this scenario I've seen
> several things I didn't quite understand.
> - A created category was displayed as a conflict, while created classes,
> methods and so forth don't. Any reason?

MC doesn't actually store category definitions, it stores organization
definitions; that is, what matters is not the individual categories but
their order.  You're not creating a category, you're modifying the order -
and this modification might conflict with another such modification.

Now, this is the kind of conflict that MC ought to be able to resolve on
its own, automatically.  There are other such conflicts, like where
only the timestamps conflict on a method.  That's on the TODO list...

> - Specifically with the RB, it seemed that some class extension methods
> appeared as differences (but not conflicts). I don't notice anything
> special about these methods. Any ideas? do you see this too?

Sorry, I'm not sure what you mean.  Can you elaborate?

Avi