Great work, Mike! I'm so excited to see this land. I'm extra excited we were able to get the experiments data included in this first batch as well. For those who don't know, this will give us data about which Switchboard experiments are active for these users. I mentioned in a bug comment [1] earlier today that it would be great to have a central place to list all experiment names and what they mean.
All experiments names live in the switchboard-experiments repo [2], so I'm going to suggest we maintain documentation there. These experiment names are also reflected in Experiments.java in the tree [3]. Be sure to update both of these places any time you add a new experiment! Margaret [1] https://bugzilla.mozilla.org/show_bug.cgi?id=1240549#c5 [2] https://github.com/mozilla-services/switchboard-experiments [3] http://hg.mozilla.org/mozilla-central/file/tip/mobile/android/base/java/org/mozilla/gecko/util/Experiments.java On Thu, Jan 28, 2016 at 4:40 PM, Michael Comella < [email protected]> wrote: > Hey again. > > It looks like the merge to mozilla-central happened late (~3am), which I > presume to be after the Nightly cutoff so we should expect to see pings to > the server and the Android documentation link in today's Nightly rather > than yesterday's. > - Mike > > On Thu, Jan 28, 2016 at 10:26 AM, Michael Comella < > [email protected]> wrote: > >> Hey all. >> >> See the subject! You should now begin seeing Android telemetry data >> hitting the servers from Nightly builds. >> >> Documentation on the Android format may be found at: >> >> https://gecko.readthedocs.org/en/latest/toolkit/components/telemetry/telemetry/core-ping.html >> >> One caveat to the current implementation is that we only try to upload >> when the application is changed from a hidden state to a partially or >> totally visible one (which includes startup; we use the `Activity.onStart` >> callback) and do not attempt to reupload the document if the upload failed >> (e.g. no network access, server is down). Because of an implementation >> quirk [1], we increment the sequence number for every attempted upload, >> even if the attempt fails. As such, users may be missing documents of >> certain sequence numbers when those fail to upload. We don't currently >> collect any data that relies on the previous document so this is >> non-consequential. This behavior is expected to be fixed in the next >> version of the telemetry implementation: >> https://bugzilla.mozilla.org/show_bug.cgi?id=1243585 >> >> What's currently landed was implemented in the following bugs: >> Implementation: https://bugzilla.mozilla.org/show_bug.cgi?id=1205835 >> Docs: https://bugzilla.mozilla.org/show_bug.cgi?id=1241599 >> A/B experiment data: >> https://bugzilla.mozilla.org/show_bug.cgi?id=1241697 >> - Mike (:mcomella) >> >> [1]: Android can kill us after we upload but before we are able to mark >> down a successful upload. To ensure we don't upload two documents with the >> same sequence, we increment for every attempted upload instead. >> > > > _______________________________________________ > mobile-firefox-dev mailing list > [email protected] > https://mail.mozilla.org/listinfo/mobile-firefox-dev > >
_______________________________________________ mobile-firefox-dev mailing list [email protected] https://mail.mozilla.org/listinfo/mobile-firefox-dev
