Yes. The first 6 hex characters of a salted md5 hash of the MAC address should be unidentifiable, and a user is unlikely to get a collision on their devices no matter how many they have.
On Sun, Aug 18, 2013 at 5:43 PM, Carl Christensen <[email protected]> wrote: > ok, I think on an android phone conference we decided MAC addresses would > involve a privacy issue - maybe we could hash it or something just to have > a quasi-unique identifier > > ------------------------------ > *From:* Eric J Korpela <[email protected]> > *To:* Carl Christensen <[email protected]> > *Cc:* "[email protected]" <[email protected]> > *Sent:* Sunday, August 18, 2013 8:13 PM > *Subject:* Re: [boinc_dev] Serious bug in BOINC android. > > Yes. Since every android machine has the same hostname "localhost", I > think any new connection by an identical device will seem as an attempt to > reattach a device. I don't know the full list of properties compared, so I > don't know if memory or OS revision is one of them or not. I hadn't > noticed it because all my hosts have different CPU revisions or a different > number of processors. > > Since I posted, I've been told there's a pending task to give android > devices the name "android-<MAC address>" where the MAC address is the WiFi > MAC address. That should solve the problem, but I'm not sure where it is > in the priority queue. > > > > On Sun, Aug 18, 2013 at 2:22 PM, Carl Christensen <[email protected]>wrote: > > is that a server-side thing where it tries to match up an unknown host > with a known host id based on the user id & hostname & cpu etc? I guess > the generic ARM naming doesn't help - I just got a Google Nexus 7 tablet > and it reports the same CPU as my last-gen Samsung Galaxy S3 phone. > > > > > > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
