I am adding glean to the reference browser here: https://github.com/mozilla-mobile/reference-browser/pull/349 Even though I have metrics disabled via setMetricsEnabled() I still see uploads happening in the log: > D Ping upload: 200 > D Ping successfully sent (200) I was expecting no upload to happen in this case?

This may be by design, I don't seem to see any information about sending or not sending the ping depending on the setMetricsEnabled() state. What I do see happening is that we are not recording any metrics if metrics are disabled, but I don't know if this affects the baseline ping because of it's lifetime attribute. I'm pinging Alessio to get more info.

(In reply to Travis Long from comment #1) > This may be by design, I don't seem to see any information about sending or > not sending the ping depending on the setMetricsEnabled() state. What I do > see happening is that we are not recording any metrics if metrics are > disabled, but I don't know if this affects the baseline ping because of it's > lifetime attribute. I'm pinging Alessio to get more info. Yes. Let's introduce a way to disable upload. This might be useful to implement the solution to the problem illustrated in bug 1514336 comment 1 as well.

Okay, by "persistently" (from the comment in bug 1514336), do we want it to be so that the glean lib manages and maintains that persistence? Since we are only a library, it seems like persistently disabling/enabling sending of pings to be the responsibility of the caller. I think having the setMetricsEnabled() flag control ping uploading is a good idea, but I'm not sure it's such a good idea to incorporate managing the persistence of that in the library itself.

(In reply to Travis Long from comment #3) > Okay, by "persistently" (from the comment in bug 1514336), do we want it to > be so that the glean lib manages and maintains that persistence? Since we > are only a library, it seems like persistently disabling/enabling sending of > pings to be the responsibility of the caller. I share the same concern about persistence, I think that should be up to the application to define. However, let's defer/take the persistence specific part to bug 1514336. > I think having the > setMetricsEnabled() flag control ping uploading is a good idea, but I'm not > sure it's such a good idea to incorporate managing the persistence of that > in the library itself. I agree: let's have this flag control everything. We can always change it back later if we need to separate the semantics between controlling recording and uploading.