[okfn-help] ORDF feedback and questions

Mon Jun 7 22:41:48 BST 2010

On 10-06-07 18:34, Alistair Turnbull wrote:
>
> As a matter of policy, I think that decision is fine (though
> debatable, as you say). However, my point is that it should be
> recorded in the change set, not applied directly to the graph.
> Constructing a patch should be a separate operation from applying it.
>

Quite right, I agree.

>> Thinking about it, you probably are right. Do you want to try
>> fixing this?
>
> Sure. I'll need write access to your repo, unless you want me to send
> you a patch?
>
> Please clarify which semantics you want. If I start from A, then apply
> diff(A, C) to obtain C, and I then compute diff(B, C), do you want to
> ignore or include A's droppings in the result? I.e. if I restart from
> B and apply the diff(B, C) I just computed, should I end up with both
> sets of droppings or just one? I'm guessing just the one set.

Basically any ORDF.changeSet present in the original graph (A or B)
should migrate to a CS.preceedingChangeSet in the changeset itself
and the only thing in C would be a pointer to the new changeset.

I'm not sure I understand the first use case - diff(A,C) diff(B,C). I
think you're thinking about merging different branches of a change
history, and that's not something I've given an excessive amount
of thought to. I think it will need some sort of special handling.
TODO future work.

Right now the only use of the changeset code is in the handler's
ChangeContext in which case A==B always and I'm not entirely
sure what would happen if you passed in different C with the
same identifier...

Cheers,
-w

-- 
William Waites           <william.waites at okfn.org>
Mob: +44 789 798 9965    Open Knowledge Foundation
Fax: +44 131 464 4948                Edinburgh, UK