On Tue, Feb 29, 2000 at 05:40:11PM +0000, Dr A V Le Blanc wrote:
> Are there any plans to implement in Coda the ability to transfer
> only 'chunks' of large files, the way AFS does?  Granted that we
> know it can be done, has anyone thought about doing it?

On Tue, Feb 29, 2000 at 01:14:29PM -0500, Jan Harkes wrote:
> Not really...
> Also when disconnected operation or conflict resolution come into play,
> it is impossible to find the complete original file once a client starts
> caching only chunks of a file. i.e. I could come back from a week long
> holiday and might not be able to reintegrate the dirty middle chunk of a
> file I had worked on somwhere during the disconnection. The other
> `clean' chuncks might have been flushed by then, and the server doesn't
> have the original anymore.
> 
> As far as I can see, it might be possible, but only if the servers keep
> a complete version history of a file.

The complication about simultaneous changes must affect all files,
surely.  The problem with reintegration could be solved by requiring
the hoard to contain all the chunks for a hoarded file.  In other
words, I can certainly see scope for problems, but it would solve
a potential shortcoming of the system: namely, that access to large
files on a connected client is rather more restricted than it
needs to be.

     -- Owen
     LeBlanc_at_mcc.ac.uk