Aside from having a preference to enable/disable Telemetry on mobile we need to advertise Telemetry to our users and encourage them to opt-in. We need some sort of UI that is exposed to the user for this purpose.

Send performance data Seems to exist in preferences for latest birch. 20111107

John, you're correct. Adding the Send performance data preference was the goal of bug 696076. This bug is about advertising Telemetry opt-in.

Comment on attachment 572988 [details] [diff] [review] patch >diff --git a/mobile/chrome/content/browser.js b/mobile/chrome/content/browser.js >+ } catch (e) { >+ // optional >+ } >+ if (!telemetryPrompted) { Add a blank line >diff --git a/mobile/locales/en-US/chrome/browser.properties b/mobile/locales/en-US/chrome/browser.properties >+# Telemetry >+telemetry.optin.message=Would you like to help improve %S by automatically reporting memory usage, performance, and responsiveness to Mozilla? Desktop uses slightly different text: http://mxr.mozilla.org/mozilla-central/source/browser/locales/en-US/chrome/browser/browser.properties#332 That is a bit too wordy for mobile IMO, but we can steal some of it. How about: Will you help improve %S by sending anonymous information about memory usage and performance to Mozilla?

We went through some iterations on the desktop string to ensure it adequately covers the types of information that we're collecting. I don't know that restricting it to memory usage and performance is correct. We may want to go more general like "Help improve %S by sending anonymous usage information to Mozilla." @sstamm - Can you please review the proposed string from a privacy perspective.

(In reply to Lawrence Mandel [:lmandel] from comment #5) > We went through some iterations on the desktop string to ensure it > adequately covers the types of information that we're collecting. I don't > know that restricting it to memory usage and performance is correct. We may > want to go more general like > > "Help improve %S by sending anonymous usage information to Mozilla." > > @sstamm - Can you please review the proposed string from a privacy > perspective. Shorter is better, so I like your suggestion. Let's get an approval on this and land it. We can always tweak the string before string freeze.

changed opt-in string

(In reply to Lawrence Mandel [:lmandel] from comment #5) > "Help improve %S by sending anonymous usage information to Mozilla." This doesn't cover everything. For example, it doesn't cover the attributes of your hardware (screen size, cpu capabilities, etc) that are being collected.

(In reply to Sid Stamm [:geekboy] from comment #8) > (In reply to Lawrence Mandel [:lmandel] from comment #5) > > "Help improve %S by sending anonymous usage information to Mozilla." > > This doesn't cover everything. For example, it doesn't cover the attributes > of your hardware (screen size, cpu capabilities, etc) that are being > collected. Suggestions? How about dropping 'usage' as in (I do remember our conversation about anonymous but it seems to me that we need some adjective in this statement): "Help improve %S by sending anonymous information to Mozilla."

(In reply to Sid Stamm [:geekboy] from comment #8) > (In reply to Lawrence Mandel [:lmandel] from comment #5) > > "Help improve %S by sending anonymous usage information to Mozilla." > > This doesn't cover everything. For example, it doesn't cover the attributes > of your hardware (screen size, cpu capabilities, etc) that are being > collected. We don't need to call everything out. We don't have that kind of space.

(In reply to Mark Finkle (:mfinkle) from comment #10) > We don't need to call everything out. We don't have that kind of space. Absolutely. I don't expect a laundry list, but we shouldn't provide a category that omits stuff. I favor "anonymous data" over "anonymous usage information", but am still brainstorming better categories. What kind of space *do* we have? Can you recommend size limits?

Let's only use the word "anonymous" if that's actually true. If users are potentially re-identifiable from the data we collect, it's pseudonymous.

rebased

http://hg.mozilla.org/projects/birch/rev/2e175acc56b9 Landed. I'll leave this bug open until we figure out the exact string we want.

(In reply to Brian Nicholson (:bnicholson) from comment #14) > http://hg.mozilla.org/projects/birch/rev/2e175acc56b9 > > Landed. I'll leave this bug open until we figure out the exact string we > want. No need. We can file a new bug for that.

I see that this prompt has been implemented as a door hanger. Is there any way to dismiss a door hanger without clicking yes or no? If so, it may make sense to follow the desktop model and implement a tri state for Telemetry. See bug 697860.

(In reply to Lawrence Mandel [:lmandel] from comment #16) > I see that this prompt has been implemented as a door hanger. Is there any > way to dismiss a door hanger without clicking yes or no? Tapping BACK button or anywhere off the doorhanger > If so, it may make > sense to follow the desktop model and implement a tri state for Telemetry. > See bug 697860. File a new bug :(

(In reply to Mark Finkle (:mfinkle) from comment #17) > (In reply to Lawrence Mandel [:lmandel] from comment #16) > > I see that this prompt has been implemented as a door hanger. Is there any > > way to dismiss a door hanger without clicking yes or no? > > Tapping BACK button or anywhere off the doorhanger Doorhanger don't actually do this yet, but they will (bug 700434).

20111114041052 http://hg.mozilla.org/projects/birch/rev/859ecdfe0168 Samsung Galaxy SII (Android 2.3.4)

(In reply to Mark Finkle (:mfinkle) from comment #17) > (In reply to Lawrence Mandel [:lmandel] from comment #16) > > I see that this prompt has been implemented as a door hanger. Is there any > > way to dismiss a door hanger without clicking yes or no? > > Tapping BACK button or anywhere off the doorhanger > > > If so, it may make > > sense to follow the desktop model and implement a tri state for Telemetry. > > See bug 697860. > > File a new bug :( Done. See bug 702281.

Filed bug 702319 to update the opt-in string.

Updated start-up test case in litmus: https://litmus.mozilla.org/show_test.cgi?id=33699