Re: dead code related to fast user switching

["William Jon McCann" <mccann jhu edu>]

On 1/18/07, Matthias Clasen <mclasen redhat com> wrote:
> On Thu, 2007-01-18 at 11:29 -0500, William Jon McCann wrote:
>
> > So, I'd really like to do some brainstorming and get some input from
> > design minded folks.
> >
> > That said, I'm a bit allergic to the idea of asking for another user's
> > password (a la darwin) within the session, for security reasons.  So,
> > I'm leaning toward something more like the windows approach.  If we go
> > this way then we may not want to have a user list on the screensaver
> > dialog at all...
> >
> > What do you think?
> >
>
> I'm not really sure what the os x/win32 approaches to user switching
> are, and how they differ from what we have. Maybe you can outline them
> a bit, as the starting point for this discussion ?

OK, I've just created a stub of a wiki page for this research:
http://live.gnome.org/GnomeScreensaver/UserSwitchingResearch

I'll be adding to it... I'd like to include very high level work-flow
descriptions as well as lower level architectural details.  Anyone
please feel free to make additions or corrections.

> The one argument I see for keeping the user list in the lock dialog is
> that it is currently very slow to get to the gdm login screen, due to
> vt switching, X server startup, and whatnot.

Yeah, that's part of what I meant when I said that our current design
is mostly dictated by design limitations in our stack - there are
many.  As much as possible, at least initially, I'd like to ignore
most of them while we brainstorm on an ideal workflow.

> Anyway, I'm currently working on a gdm patch to make it possible to
> pick the user in the lock dialog/applet, and then have the login screen
> come up asking for the password. The current situation where you pick
> a user in the lock dialog and then have to enter the username again
> in the lock dialog is just absurd.

That's awesome.  Thanks for doing that.  GDM is a big piece of this
puzzle.  Depending on what kind of user switching workflow we want we
may have to make some significant changes.

We should also start thinking about how to manage the "who" part of
user switching.  The way we handle user information, including but not
limited to face images, is pretty weak.

Jon