Sounds to me like optimistic concurrency control is appropriate here -
start, and probably there will be no big clash. 

Especially since something similar to the current interface will
probably be useful in the short term anyway, even if it becomes just a
temporary facade to a smarter mechanism later. 
BTW, what're the next goals KCP is interested in?

Daniel

Andreas Raab <andreas.raab at gmx.de> wrote:
> > After discussing with Stef we saw that we did not understand at first 
> > what you meant by moving target. I think you mean: 'probably somebody 
> > else is going to touch this SourceFiles mess in order to have a 
> > different source pointer incoding, so don't you guys mess 
> > with this as well'. True?
> 
> Yes, that's the first half of it. The second is that it's likely that if
> somebody plays with it, SourceFiles aren't going to survive anyway. It's
> just an odd construct to begin with.
> 
> > If you do, my next obvious questions is: is somebody going 
> > to do this now? If not (or only after a while), I do the changes I 
> > proposed before (because it hurts every time I see them). If so, then 
> > be my guest and change this ugly looking code as well. So, 
> > what will it be?
> 
> How should I know? ;-) I was saying that I'm not going to mess with it but
> surely I can't speak for anyone else. Craig is certainly doing some work in
> this area, Avi expressed some interest but I can't tell you if (or what) is
> going to happen soon or not.
> 
> Cheers,
>   - Andreas