On 20.08.2011 17:13, Rob Weir wrote: > Existing OpenOffice.org prompts user after install on whether they > wish to participate in the "OpenOffice.org Improvement Program". The > dialog says, "If you are willing to participate, we will collect > anonymous statistics about how you use OpenOffice.org. This data is > used to identify usage patterns". > > A few points on this: > > > 1) Where does this data live today? What server?
IIRC it was collected in the "Hamburg lab". > 2) This data likely has some value to the project. Is it coming to > Apache? If so, we need a plan for determining whether this data needs > to be treated as sensitive. For example, did it collect IP addresses? No, we collected user interactions: which functionality was used, was it used via menu or toolbar, which buttons in a dialog have been pressed and in which order etc. No IP addresses or anything else that could be used to create relationships between data sets from the same user. IMHO it doesn't make sense to continue collecting data as long as nobody is interested in working with it. In the old project it was our "user experience" team. We have to see if something comparable will materialize inside the AOOo project. > 3) As part of the transition to Apache, will the collection point for > this data be moving to a different address? If so, what effect does > that have on legacy installations of OOo? Will it slow them down? > Cause a time out error for the user? Or is it handled silently and > with no user impact? Whatever we will do, the old installations will no longer find a server. The data is collected and written to disk (as a csv file that the user easily can load into Calc to see what is collected), and in certain intervals OOo tries to send the file to the server. An attempt to send the data will not block the application, so I don't expect to see a slowdown if the server is shut down. It also won't disturb the user with error messages. Of course the application also should stop collecting data (otherwise it would just fill the hard disk with). I don't remember exactly when OOo will do so (perhaps after n futile attempts to send data). I would be surprised if we hadn't planned for this case, but I fail to remember the exact details. Regards, Mathias
