High level overview of future plans/ideas. There is no set schedule and everything listed here is up for debate, refinement, addition, or could be dropped.
If you have feedback or anything you'd like to see here, please start a thread on the mailing list.
The below are in no particular order and will likely lack implementation details.
It's pretty hard to get off the ground with MogileFS. First you set up the server system, then your application needs to be tuned to look up and reproxy files. A perlbal plugin should be provided with the distribution which is able to directly serve files itself.
A preview of working code has been posted to the mailing list: http://groups.google.com/group/mogile/browse_thread/thread/e3a33161d2da1d2a/00889fe6893fab21
Some or all of the wiki should be moved into POD docs in the code, and then the wiki generated off of that. Or at very least, a lot more of it needs to be documented.
A lot of the times we tell people to telnet to a tracker and run some
commands. Instead, mogadm should be able to centrally control jobs,
monitoring, etc. More of the tracker configuration should be moved out of
config files and into the database.
This feature is implemented in 2.60 and refined in later releases. See Checksums.
The protocol is flat and very manual. Replacing it with a JSON based protocol would allow a huge simplification protocol handling code. Many languages have JSON libraries available by default now, so it would also make it simpler to port.
Replication should be traceable, verifyable, and have a much better extension API. Currently "ran out of ideas for replicating blah" is often the most information you'll get.
It's possible to add flexibility for how mogile handles deletes and updloads. All changes could be replicated to a backup system, S3, etc.
It should be possible for all overwritten files to be versioned via a history table. Making it innately possible to undo deletes or seek around in history, useful for a backup system.