ubuntuone-users team mailing list archive

[Rodney Dawes <rodney.dawes@xxxxxxxxxxxxx>]

On Thu, 03 Dec 2009 14:03:58 +0800
Simon Wiles 魏希明 <simonjwiles@xxxxxxxxx> wrote:

> Hi Elliot, Rodney, thanks for your replies.
> 
> It seems the issue may be more complicated that I imagined.
> 
> I don't understand this part, though:
> 
> On Wed, 2009-12-02 at 00:18 -0500, Rodney Dawes wrote:
> > Once we switch to using python-keyring to abstract the keyring bits
> > out, this won't actually be an issue any more. Of course, the other
> > major problem is that we still need UI to do the authentication
> > step. Because u1sync has options to pass in the oauth token and
> > consumer information, it still allows for authenticating if there
> > isn't a token in the keyring. I'd rather not remove that
> > functionality. At that point,
> > supporting headless setups is a bit more problematic. 
> 
> Doesn't passing the oauth token obviate the need for the
> authentication step?  Why would anyone want to remove that?  And
> doesn't it make operating in a headless environment *easier*?

Yes and no. Passing the token as an argument is doable. But it still
requires that the keyring dependencies be installed. If they were to be
removed, you would get an ImportError. The plan is to have more robust
authentication service for Lucid, which should make headless a lot
easier to deal with.

> Please consider operation in a cli-only environment my submission to a
> wish-list, then.

You might want to subscribe to the following bug, requesting headless
support: https://launchpad.net/bugs/440351

Thanks.