Bradley Kuhn on switching back from Ubuntu to Debian

Elliot Murphy elliot at
Mon Jan 18 22:44:39 GMT 2010


On Sat, Jan 16, 2010 at 6:21 PM, Neil Broadley <scaine at> wrote:
> I think that there would be considerably less opposition/vitriol regarding
> Ubuntu One if it were to offer the ability to use a number of different
> back-end services.  At the moment, it will only sync with the Amazon service
> rebranded by Canonical as UbuntuOne, but perhaps in the future, it will
> allow you to sync to any publicly available service - dropbox, humyo, moxy,
> etc.  Of course, they are all closed-source too, but the way the cloud is
> shaping up, perhaps there's room for a free player in the future.  For
> example, how difficult might it be for the Ubuntu One client to support
> webdav?  That would open a few doors.

I don't currently have a great solution for the fact that I and
Canonical have not released the source code to the Ubuntu One file
sharing server. To be clear, it's not a rebranded Amazon service but
is a file system written on top of S3. I and the other people working
on Ubuntu One did take the objections very seriously, and so I wanted
to tell you a bit about how we built the next phase of Ubuntu One
after file sync - contact syncing, bookmark syncing, and Tomboy Note

For Tomboy Notes, we collaborated with Tomboy upstream, helping them
improve their REST API, and even contributing code to Snowy, the AGPL
server for Tomboy note sync. We then implemented that same API on For contacts and bookmarks, we built on top of
CouchDB, and run a CouchDB cluster in the cloud. This means that
applications which integrate with Ubuntu One continue to work even if
the user doesn't have an Ubuntu One account, the user can set up their
own couchdb server and get the cool cloud-enabled network syncing
features, and if you want to use Ubuntu One it all just works (thats
the theory, there are bugs of course). I'm quite proud of this design,
and the way it preserves user autonomy while also meeting our goals of
an Ubuntu One service that "just works" for ordinary users.

Whenever we are discussing Ubuntu One features, I advocate a design
principle for Ubuntu One: "build no silos". I'm not pretending that
the Ubuntu One file sharing server source code being closed is a
wonderful thing, but I do believe that the end user impact if Ubuntu
One file sharing suddenly disappears is not so bad. The user has all
their files, and can continue working. The switching cost to dropbox
is very very low. So it's not Franklin Street Free at the moment, but
neither is it a silo.

I've made public offers before to offer advice and coaching if someone
wanted to implement a simple server that speaks the Ubuntu One file
storage protocol. Interestingly, I've gotten some feedback from
developers saying they didn't want to implement the protocol because
it used AGPL, and they wanted to make closed source implementations -
but no takers on a free implementation. I've been trying to figure out
what it would take to split out some useful file sharing server code
to let people run their own Ubuntu One server at home if they wanted
to without all the dependencies on S3 and EC2. It's not as easy as I
would like, but I haven't stopped thinking about it, and maybe one day
I'll have good news there too :)

My point is simply that we're listening to the criticisms, and are
constantly trying to adjust course as best we can. I'm quite excited
to be working on this service, and I hope that it's able to contribute
to ongoing funding of Ubuntu development.
Elliot Murphy |

More information about the ubuntu-devel mailing list