> 
> I still like the idea a lot, I'm just trying to be prepared and know all
> the problems that might show up before actually trying to implement
> something like this.
> 

I think a lot of the issues that got brought up will be handled by David
Howell's cachefs.. My thoughts are the least intrusive way to handle
streaming reads is to let cachefs do the hard deadlock-prone stuff. It's
going to get more tested since it's likely going to be the backend data
store for both the in-kernel AFS client and probably used as a backend
for NFS as well. Then we can keep Coda focused on the disconnected
operation problem instead of getting lost in cache ;)