Has anyone considered splitting off the persistent caching of CODA into
a more general purpose caching mechanism and have CODA sit on top of that?  Any
slow filesystem could benefit from a reliable, local harddrive cache.  The
application I have in mind is actually rather interesting.  I would like to
modify CODA to cache data from a CDROM jukebox instead of a network server.

	If I understand it correctly, currently when CODA gets a request for
information on the filesystem, it sends that request to a user-space damon.
This damon checks the local harddrive cache and upon failing there, goes to
theconfigured CODA server for the information.  The cache gets updated and the
kernel gets notified that the data is available.  How hard would it be to
modify this process to talk to a CDROM jukebox instead of the CODA server?  Is
it as simple as writing a new class to replace a network communication class
with a class that moved disks around and mounted cd's as requested?  Any
pointers as to where to start looking in the code would be greatly appreciated.
The fact that there seems to be around 300,000 lines of source seems a little
intimidating.

By the way, this jukebox has 100's of disks and many CDROM drives.  Persistent
caching is the only sane way to make this work in a reasonable way.

Thanks,

Aaron