On Tue 24 Jun 2008 at 06:32PM, Christopher Kampmeier wrote: > It seems like even IPS itself could sense movement of an IPS image and > force a UUID reset or unset automatically. Failing that, image > management tools such as an installer could force a reset.
The definition of "movement" is tricky. Unless you want to do something icky like compare inodes, I'm not sure how to do this. Maybe I'm overlooking something obvious. > >One of the comments that Stephen supplied in this thread was that the > >>UUID should be opt-out, not opt-in. > > > I'm expecting that this will be handled by an installer. A UUID cannot > be set for a pre-installed image that is copied from a CD or downloaded > because then everyone would have the same one. So it has to be set by > the installer (by calling --reset-uuid). That part will be automatic, > and to opt out, someone can run --unset-uuid. I was wondering why we'd do this by setting and unsetting the UUIDs in this way-- let's just always maintain a UUID for every image, and then let the user select whether that UUID is transmitted or not. As an aside: I would like the design to be flexible with respect to opt-in-by-default or opt-out-by-default, in case we need to alter that setting in the future. I passed this along to Stephen, and will share it here: http://john.jubjubs.net/2008/05/13/mozilla-firefox-data/ (see especially the comments). The redux is that different people in different parts of the world have different opinions about anonymity and privacy. We may think those terms are either fairly black and white, or fairly static, but they aren't. Staying flexible here may be important. -dp -- Daniel Price - Solaris Kernel Engineering - [EMAIL PROTECTED] - blogs.sun.com/dp _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
