On 6/22/10, Pappa <[email protected]> wrote:
> Okay...  I was forced to change my email address from the one I used since
> Boinc Started (2005) when I set it up and really started using the Seti
> Boinc Account. What then happened is somewhat close to CHAOS. Obviously the
> email hash is going to change.. But WHY does the Cross Project Identifier
> need to be changed in the production of stats output?
>
> Old Cross Project ID. 71019c79ef524efd9af21814073ffaf0
>
> New Cross Project ID. 96ce42b2d871123340db773a20707e32
>
> If I look in any of my sched_reply_someproject.xml I come up with something
> different
>
> <email_hash>3de9749c4e6a16c91875bb4264111446</email_hash>
> <cross_project_id>72bfb0ef0cc023be98f1c1ddcd4355a6</cross_project_id>
>
> I find nothing in common to correlate anything with what is in my Boinc
> Folder. There is only PROOF that it HAPPENS!

The "public" CPID is a hash of the "private" CPID (the one you see in
sched_reply) and your email address.

This is to prevent impersonating other people. Otherwise I could
change my client_state.xml to have *your* CPID. With this hashing, if
I change my private CPID to 72bfb0ef0cc023be98f1c1ddcd4355a6, the
server will hash it with my email address and I'll have a different
public CPID than you.

The whole mechanism is described here:
http://boinc.berkeley.edu/wiki/Cross-project_identification

> Once a user is established, there is no reason to change the Cross Project
> ID. It is hard enough to "sync it" when adding a new project. One would
> almost have to think it "should" track email hash_old + Cross Project ID
> and the add hash_new + the same Cross Project ID. The use tables could be
> updated and then projects that could not be contacted (down) would make
> little difference.

Keeping track of old CPIDs sounds like a good idea...

-- 
Nicolas
_______________________________________________
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.

Reply via email to