GitLab

4693: Fennec data migration scaffolding r=pocmo a=grigoryk

This is going to be a rolling PR for a little bit, until this is ready to land.
Companion Fenix PR: https://github.com/mozilla-mobile/fenix/pull/5878

Workflow for testing it:
- fetch an ESR68 build from TC: https://tools.taskcluster.net/index/gecko.v2.mozilla-esr68.signed-nightly.nightly.latest.mobile
- re-sign it using a local debug keystore:
  - `zip -d target.apk "META-INF/*"`
  - `~/Code/mozilla-central/mobile/android/debug_sign_tool.py target.apk`
- enable auto-publication workflow for a-c in fenix
- assemble fennecProduction version of Fenix (built against this branch, see auto-publication)
  - `./gradlew assembleFennecProduction`
- sign it using keys used to sign fennec esr apk (otherwise we won't be able to overwrite one with another):
  - `~/Code/mozilla-central/mobile/android/debug_sign_tool.py app/build/outputs/apk/geckoBeta/fennecProduction/app-geckoBeta-x86-fennecProduction-unsigned.apk`
- install fennec esr apk build
- generate some profile data for migration (sign-into fxa, browse around, etc..)
- install fenix build: `adb install -r app/build/outputs/apk/geckoBeta/fennecProduction/app-geckoBeta-x86-fennecProduction-unsigned.apk`

TODO:
- currently needs https://github.com/mozilla/application-services/pull/1942

 to work
- add separate migration versioning for history and bookmarks; currently the whole thing is versioned
- add migration tests
  - in fennec, we have a bunch of "schema migration" tests that operate over older versions of browser.db
  - this is a good approach here as well - we can run migration code in tests against a few test db files, that exercise various history/bookmarks configurations and try multiple schema versions
- improve the workflow around this, it's quite tedious currently
Co-authored-by: Grisha Kruglov <gkruglov@mozilla.com>

4811: Issue #4810: browser-state: Child tabs not added next to their parent r=pocmo a=csadilek

This is a follow-up to https://github.com/mozilla-mobile/android-components/pull/4798

.
Co-authored-by: Christian Sadilek <christian.sadilek@gmail.com>

4798: Set parentId to session before adding tab r=csadilek a=mawen7
Co-authored-by: mawen7 <mawen7@users.noreply.github.com>

4789: Use interface with PromptDialogFragment r=Amejia481 a=NotWoods

Created interface for `onCancel`, `onConfirm`, `onClear`
Co-authored-by: Tiger Oakes <toakes@mozilla.com>

4773: Closes #4749: Fix for bug 1522829 r=pocmo a=csadilek

We basically already have the parent session everywhere we need it, we now just forward the corresponding engine session so we can pass it along to GeckoView e.g. when opening a new (regular or private) tab, and when reacting to window requests.

@pocmo one discussion point is whether or not we have to create an engine session for the parent if it's no longer there (e.g. when re-loading a child tab after a restart). I decided **not** to do that as we'd be loading all parent sessions on startup and the child session should already have the right state.

Also, I didn't want to overload the method as named args with default values make this easy enough to read in Kotlin. 
Co-authored-by: Christian Sadilek <christian.sadilek@gmail.com>

4807:  GeckoView update (nightly) (20191021-142148) r=Amejia481 a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>

4788: Abstract away activity/fragment in prompt feature r=Amejia481 a=NotWoods
Co-authored-by: Tiger Oakes <toakes@mozilla.com>

4801:  GeckoView update (nightly) (20191020-141037) r=pocmo a=MickeyMoz



4802:  Docs update (20191021-121133) [ci skip] r=pocmo a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>

4786: Fix the Glean instrumented tests r=Dexterp37 a=Dexterp37

The tests were mistakenly re-initializing Glean when running each test, in order to send pings to
a local pingserver. However this is problematic as instrumented tests run as a separate thread in the
application process and Glean now asserts if it is initialized off the main thread.
Co-authored-by: Alessio Placitelli <alessio.placitelli@gmail.com>

4739: Closes #3720: Incorrect selection notification when session removed r=pocmo,rocketsroger a=csadilek

This fixes two issues: 

- The issue described in #3720 where removing the selected session does not cause a `onSessionSelected` notification

- Sending an unnecessary `onSessionSelected` when removing a session to the left of the selected session. In that case the index changed but it's still the same session that's selected and observers should not need to get notified that a session was selected
Co-authored-by: Christian Sadilek <christian.sadilek@gmail.com>

4784: Closes #4783: service-experiments staging endpoint is incorrect r=pocmo a=travis79
Co-authored-by: Travis Long <tlong@mozilla.com>

4782:  Docs update (20191018-122202) [ci skip] r=Amejia481 a=MickeyMoz



4785:  GeckoView update (nightly) (20191018-141230) r=Amejia481 a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>

The tests were mistakenly re-initializing Glean
when running each test, in order to send pings to
a local pingserver. However this is problematic as
instrumented tests run as a separate thread in the
application process and Glean now asserts if it
is initialized off the main thread.

4767:  GeckoView update (nightly) (20191017-141848) r=pocmo a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>

4762: For #4761 - Adds setting to control `forceUserScalableEnabled` r=csadilek a=boek



4769: For #2923: Don't crash if Intl.RelativeTimeFormat not supported r=Amejia481 a=csadilek

`Intl.RelativeTimeFormat` is in Nightly and Beta, but not yet on Release (until Monday). Not a problem as Fenix is on Beta (and R-B on Nightly) anyway, but good to catch errors anyway. This would also help if an article specified an invalid lang for instance.
Co-authored-by: Jeff Boek <jeff@jeffboek.com>
Co-authored-by: Christian Sadilek <christian.sadilek@gmail.com>

4746:  GeckoView update (nightly) (20191016-142200) r=pocmo a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>

4745: Fail complete tasks faster so bors errors out faster r=pocmo,mitchhenges a=JohanLorenzo
Co-authored-by: Johan Lorenzo <jlorenzo@mozilla.com>

4760: Closes #2923: Remove moment.js from reader view web extension r=Amejia481 a=csadilek

Removes moment.js from reader view, and uses [Intl.RelativeTimeFormat](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RelativeTimeFormat

) instead to construct a localized reading time string.

Tested this with various sites in Spanish, Arabic, German, and English to verify this works as before. We could improve this more later to detect the language if the article doesn't specify but that's a separate issue (which also doesn't work in Fennec currently).
Co-authored-by: Christian Sadilek <christian.sadilek@gmail.com>

4744:  Docs update (20191016-121549) [ci skip] r=pocmo a=MickeyMoz
Co-authored-by: MickeyMoz <sebastian@mozilla.com>