[Cosmo-dev] Interesting idea to avoid EIM conflicts

[Phillip J. Eby]

At 02:22 PM 2/13/2007 -0800, Jeffrey Harris wrote:
>Hi Morgen,
>
> >> In the interim state where we're displaying the item with a conflict, it
> >> seems like Arthur's Chandler ought to say the item was changed by
> >> Arthur, but has conflicts from Bea.
> >
> > Actually, there is a big problem with this: what if Arthur changed the
> > title, then Charlie changes the title and syncs, then Bea changes the
> > body and syncs, then Arthur syncs.  Arthur sees a conflict on title.
> > But the conflict isn't with Bea, it's with Charlie!  Bea just happened
> > to be the last person to update the item.
>
>Yup.  The original goal was to track who changed what attribute.  The UI
>for that is a little harder than just a lastModifiedBy field, though, so
>it's not on our Preview plate.
>
>For preview, I think it's OK if we're inaccurate about who exactly the
>conflict is with.  For now, since Bea made the last change, the conflict
>appears to be with her.  I believe Mimi explicitly blessed that
>inaccuracy for now (cc:ing Mimi to double check that).

Note that knowing who made the last change is not the same as knowing who 
you have a conflict with.  The approach I proposed (of simply keeping the 
most-recent last-modified-by user+date) actually ensures that we always 
have a correct answer for "who last changed the record" -- *whether their 
changes were applied or not*.

So, by that definition, the conflict-free approach is 100% correct all the 
time.  Whether that's actually what we *want* is a separate question.  :)