The groundwork is in place from Bug 1119520 and dependents. This ticket tracks trying to flip --with-gradle for Nightly builds, and eventually removing https://dxr.mozilla.org/mozilla-central/source/config/makefiles/java-build.mk and everything that consumes it entirely.

In the wild, the Robotium Maven dependency is not being fetched. Let's see if this forces it to be downloaded. Review commit: https://reviewboard.mozilla.org/r/40573/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40573/

Review commit: https://reviewboard.mozilla.org/r/40575/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40575/

Review commit: https://reviewboard.mozilla.org/r/40577/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40577/

This permission was added in API 16, and is only enforced in API 19+. (It's benign to add it to APKs installed on API 15.) We want to declare it explicitly so that the bouncer APK and the main APK have the same permission set. There appears to be some fanciness with Gradle's implied permission system where the bouncer APK does not request READ implicitly where-as the main APK does request READ implicitly. This just makes things explicit (and uniform). Review commit: https://reviewboard.mozilla.org/r/40579/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40579/

Review commit: https://reviewboard.mozilla.org/r/40581/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40581/

Review commit: https://reviewboard.mozilla.org/r/40583/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40583/

Gradle produces signed, unaligned APK files. We expect unsigned, unaligned APK files. This change discards any existing signature, turning a signed, unaligned APK into an unsigned, unaligned APK. Review commit: https://reviewboard.mozilla.org/r/40585/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/40585/

Comment on attachment 8731420 [details] MozReview Request: Bug 1254353 - Pre: Add GRADLE_FLAGS to control Gradle in automation. r?chmanchester https://reviewboard.mozilla.org/r/40577/#review37327 ::: old-configure.in:4703 (Diff revision 1) > if test -z "$GRADLE" -o ! -x "$GRADLE" ; then > AC_MSG_ERROR([The program gradlew/gradle was not found. Use --with-gradle=/path/to/bin/gradle}]) > fi > fi > AC_SUBST(GRADLE) > +AC_SUBST(GRADLE_FLAGS) Surely Python configure can handle this. We have a mobile/android/moz.configure. If that's giving you trouble let me know and I'll come back with a patch.

Comment on attachment 8731421 [details] MozReview Request: Bug 1254353 - Pre: Explicitly request READ_EXTERNAL_STORAGE permission. r=sebastian https://reviewboard.mozilla.org/r/40579/#review37583 The commit message says "_MANIFEST": READ_EXTERNAL_MANIFEST

Comment on attachment 8731418 [details] MozReview Request: Bug 1254353 - Pre: Assemble app and tests when collecting dependencies. r=me Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40573/diff/1-2/

Comment on attachment 8731421 [details] MozReview Request: Bug 1254353 - Pre: Explicitly request READ_EXTERNAL_STORAGE permission. r=sebastian Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40579/diff/1-2/

Comment on attachment 8731424 [details] MozReview Request: Bug 1254353 - Pre: Allow re-signing already signed APK files. r=gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40585/diff/1-2/

Comment on attachment 8731423 [details] MozReview Request: Bug 1254353 - Include Robocop support files. r?gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40583/diff/1-2/

Comment on attachment 8731419 [details] MozReview Request: Bug 1254353 - Build with Gradle in automation. r?gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40575/diff/1-2/

Comment on attachment 8731424 [details] MozReview Request: Bug 1254353 - Pre: Allow re-signing already signed APK files. r=gps https://reviewboard.mozilla.org/r/40585/#review37973 Please add an inline comment basically stating what you said in the commit message because the source code is not obvious.

Comment on attachment 8731423 [details] MozReview Request: Bug 1254353 - Include Robocop support files. r?gps https://reviewboard.mozilla.org/r/40583/#review37975 ::: mobile/android/tests/browser/moz.build:12 (Diff revision 2) > - 'robocop/roboextender', > + 'robocop/roboextender', > -] > + ] > +else: > + TEST_DIRS += [ > + 'robocop/roboextender', > + ] Having robocop/roboextender listed twice is weird. It should be pulled outside of a conditional.

Comment on attachment 8731419 [details] MozReview Request: Bug 1254353 - Build with Gradle in automation. r?gps https://reviewboard.mozilla.org/r/40575/#review37977 How were the archives uploaded to tooltool produced? What happens if you get hit by a bus? How can we reproduce those archives? Are the archives produced deterministically? We should ideally have tools in tree to reproduce the archives (if they aren't already). See build/windows_toolchain.py for what I recently wrote for VS2015. I also added docs under build/docs describing how to run it. We need all this stuff documented and as reproducible as possible.

Comment on attachment 8731424 [details] MozReview Request: Bug 1254353 - Pre: Allow re-signing already signed APK files. r=gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40585/diff/2-3/

Review commit: https://reviewboard.mozilla.org/r/41627/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/41627/

Comment on attachment 8731423 [details] MozReview Request: Bug 1254353 - Include Robocop support files. r?gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40583/diff/2-3/

Comment on attachment 8731419 [details] MozReview Request: Bug 1254353 - Build with Gradle in automation. r?gps Review request updated; see interdiff: https://reviewboard.mozilla.org/r/40575/diff/2-3/

Comment on attachment 8733187 [details] MozReview Request: Bug 1254353 - Pre: Document how to produce Android Gradle dependencies. r?gps https://reviewboard.mozilla.org/r/41627/#review38053 This is awesome. Thank you.

Comment on attachment 8731423 [details] MozReview Request: Bug 1254353 - Include Robocop support files. r?gps https://reviewboard.mozilla.org/r/40583/#review38057

Comment on attachment 8731419 [details] MozReview Request: Bug 1254353 - Build with Gradle in automation. r?gps https://reviewboard.mozilla.org/r/40575/#review38059

During eng meeting, we were discussing how having an all-gradle build could make Photon work easier by allowing a different Gradle config for Photon work (slated to ship in 57?).

(In reply to Chenxia Liu [:liuche] from comment #33) > During eng meeting, we were discussing how having an all-gradle build could > make Photon work easier by allowing a different Gradle config for Photon > work (slated to ship in 57?). My expectation is that you could develop Photon entirely --with-gradle using a different Gradle flavour, and then "flip the switch" and push all the new things into the existing moz.build. That would achieve the separation you want without requiring us to solve the outstanding problems blocking building Nightly and Release --with-gradle. Consulting https://wiki.mozilla.org/RapidRelease/Calendar, I see that Fennec 57 opens in June -- 2 months out. Given the tight timeline, if the Fennec team wants to push for --with-gradle builds for Photon, we need to talk now about my availability. I happen to have picked up bits of this last week just for fun, and I am happy to help others actually make this happen, but I think that I will have to be heavily involved for this to succeed.

Let me redirect this question to Barbara/Wesley: Is Android Photon work slated to ship in 57 along with the Desktop ship of Photon? It wasn't clear during our Android eng meeting if that was the plan, although it looks like there is a Photon work week in YVR the week of May 8.

(In reply to Chenxia Liu [:liuche] from comment #35) > Let me redirect this question to Barbara/Wesley: > > Is Android Photon work slated to ship in 57 along with the Desktop ship of > Photon? It wasn't clear during our Android eng meeting if that was the plan, > although it looks like there is a Photon work week in YVR the week of May 8. Yes. We are in the progress of scoping Photon on Fennec, targeting 57. Likely will focus one visual refresh and animation changes those potentially improve perceived performance.

Yes, Photon Mobile *must* ship with Photon on Desktop.

(In reply to Barbara Bermes [:barbara] - NI please! from comment #37) > Yes, Photon Mobile *must* ship with Photon on Desktop. OK, let's take discussion of what's needed technically for Photon Mobile to a ticket. However, I can't find a candidate ticket! barbara, sebastian: where can we discuss what's actually needed from the build system to support Photon Mobile? (This ticket is a meta ticket and tracks a different project, so it's not a good place for that discussion.)

I need to talk to antlam once he's back next week to know about the scope of Photon to answer that. Right now I was merely brainstorming ideas about how to land disabled bits of Photon in 55/56 - without blowing up the APK size. Gradle module overlays seemed to be one promising approach. I'll make sure to file tickets once I know about the scope.

I just filed Bug 1355774 as a meta to track Photon Mobile on Firefox for Android. For the time being (before we have bugs for each specific bugs) I think it's fine to have all kinds of discussions going on that meta.

Making a note of this for myself: per https://dxr.mozilla.org/mozilla-central/rev/c697e756f738ce37abc56f31bfbc48f55625d617/mobile/android/tests/browser/junit3/src/org/mozilla/tests/browser/junit3/TestJarReader.java#77, we need to stuff package-name.txt into Gradle-produced APK files.

The transition to Gradle is finally going to happen \o/ I'm going to split this into v1 (now) and v2 (future) meta tickets, and move them out of Core :: Build Config, since they're really Firefox for Android-specific.

Re-triaging per https://bugzilla.mozilla.org/show_bug.cgi?id=1473195 Needinfo :susheel if you think this bug should be re-triaged.