Below is a first-pass response on some of the more simple parts of the thread. Later, I'll follow up on the remaining bits in one or more emails.

Jared Rhine wrote:
One next step I'll take is to cut-and-paste your list as-is into a new PPD dashboard page.

Ok, so I haven't gotten very far, but I've created the page, and adding a few of the simplest stats I have so far. See http://dashboard.osafoundation.org/dashboard/ppd , where I'll add more stuff on this thread over time.

# of unique user accounts
# of items
# of subscriptions

These are now available on the page, along with # of events (a subset of items).

Do each of the bars in these graphs represent totals? As in, the Total # of items/events/subscriptions and tickets on the server?

Yes, they are totals. Technically they are exactly the number of rows in the table named that thing in the Cosmo relational database schema. But they accurately reflect the concepts above.

Let's spec out "unique publishers" and "unique subscribers" more; is that "total number of people who have published a collection"?

+ unique publishers, as in unique Hub accounts?

Isn't this equal to "# of unique user accounts" as above?

+ unique subscribers is trickier as it involves anonymous subscriptions from the Desktop. So #of desktop clients with subscriptions?

Number of Desktop clients who are subscribing to a collection via a ticketed URL? I do believe that one should be straightforward to extract from our current logs. Given a (munged) real-world log entry:

201.78.226.17 - ticket:wguqp77770 [03/Dec/2007:00:04:32 -0800] "GET / mc/collection/6cc87777-6eb6-1177-fdfe-e3864cd77777 HTTP/1.1" 200 153 "-" "Chandler/0.7.0.1 (Windows; U; i386; pt_BR)"

I can count up how many of the Desktop users (people with "Chandler/*" as the user agent) access at least one URL with "/mc/collection" with a "ticket:*" in the 3rd field (username/ticket). This count will be per day like other metrics. This number will be a subset of the "total Desktop users by day" currently given by the dark-blue segment of the top graph here: http://dashboard.osafoundation.org/

That dark-blue segment is the total unique "Desktop users" seen each day (where unique is defined by that "unique IP+user-agent pairs" outlined in my first email). So of that total number (120 on a good day), I can break that into two parts: those using at least 1 ticket, and those not using any tickets.

That's the most accurate technical implementation I can translate the above item ("# of desktop clients with subscriptions") as I understand it. Does that suit?

"Unique subscribers" is "number of accounts which have any subscriptions created"?

# of accounts on Hub + anonymous subscribers from the Desktop + iCal subscribers, etc...

"# of accounts on Hub" means to me "total number of accounts created on the Hub", or the "# of unique user accounts" graph (from the database table) at the top of the new PPD dashboard. So the above 3 categories will be greater than oh, the 3400-odd that "# of unique users" graph stands at today?

Hrm, well I guess it'd be nice to figure out if a single person has multiple accounts...

I can't think of any realistic way to do this.

Looking back over this list, I think the highest priority for me is to figure out a way to break down our numbers so we have numbers for Desktop users syncing to the Hub.

As I understand that concept, it is closely matched by the dark-blue segment mentioned above on the "Daily Hub Visitors By Application" graph. I've split that number off and placed it second on the PPD dashboard as "# of Desktop users syncing to the Hub".

-- Jared

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "General" mailing list
http://lists.osafoundation.org/mailman/listinfo/general

Reply via email to