[Design] Interesting idea to avoid EIM conflicts

[Phillip J. Eby]

At 02:19 PM 2/14/2007 -0800, Mimi Yin wrote:
>+ Byline value doesn't change if there are conflicts and the
>conflicting changes haven't actually been applied

That sounds good, except that it's not really defined.  In the edit/update 
case, I could email you a change Morgen made, along with a change that I 
made previously (but Morgen's change is later, so my lastModifiedBy says 
Morgen).

Now, if one or both changes conflict with your Chandler, then what do we do 
to your lastModifiedBy?  Under the model you propose, it is quite possible 
to have a situation where, no matter what we do, the value will be wrong!

The only way that I see to have a model that is always "right" (by 
construction) is to make lastModifiedBy always simply reflect the user who 
has most recently attempted to change the item, regardless of whether their 
changes are present.

This model can be made 100% consistent with ease, and it actually provides 
useful information, as you at least then know who is most likely the best 
person to contact regarding the conflict.

In addition, it trivially satisfies the "everyone sharing sees the same 
value" requirement, because by its very definition it can *never* be in 
conflict, and thus is never subject to unapplied changes.