On Friday, 9 March 2018 13:31:18 CET Aleksey Kontsevich wrote: > Hi Volker, > > Have some questions. I've read Telemetry discussion in kde-community mailing > list, however some KUserFeedback logic is not quite clear: > https://firstname.lastname@example.org/msg03898.html > > You said data is submitted once per week to a server. How the user is > identified as I think the returned statistics contains data from previous > sessions, but there is no user identification in KUserFeedback? > > We thought to add at least machine ID like Krita does. Identifying the user > is important. If the frontend sends the data once a week, we still want to > know, how design mode usage changes between the weeks. If developer has > several installations in different hosts, it would be sufficient to know > installation-specific statistics. > > Could You clarify our questions and concerns, please? Does returned > statistics contain data from previous sessions or not? If aggregation > performed only in analytics tool and server contains raw data how > statistics integrity is provided? How data collected between submissions, > does statistics contain data from previous submission or not? Does it makes > sense to add some randomly generated machine ID on first connection to the > server for satisfy our needs? Other thoughts?
I've added some more details about this to the docs: https://commits.kde.org/kuserfeedback/4c685a70afcf6ba58ce4455e88ef339c6d9e37b8 As you are not bound by the KDE telemetry policy you can add a custom data source with a unique id, that should be straightforward to implement. Historic data is by default not submitted, but custom data source have the ability to build a history locally, and submit that in whatever form they want. Regards, Volker
Description: This is a digitally signed message part.