1237755 - adb shell am start ... no longer actually opens links if Fennec is already running

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: General, defect)

Description

[Nick Alexander :nalexander [he/him]]

I see:

$ adb shell am start -a android.activity.MAIN -n org.mozilla.fennec_nalexander/org.mozilla.gecko.BrowserApp -d 'http://people.mozilla.org/~ewong2/push-notification-test/'
Starting: Intent { act=android.activity.MAIN dat=http://people.mozilla.org/~ewong2/push-notification-test/ cmp=org.mozilla.fennec_nalexander/org.mozilla.gecko.BrowserApp }
Warning: Activity not started, its current task has been brought to the front

But nothing happens.  Something in the Intent handling code?  The Tab Queuing code?

If Fennec isn't running, this works.  It's just when Fennec is already running that this is ignored.  I've just tested with a fresh install before I've seen the Tab Queuing prompt, and it's busted.

Comment 1

[Nick Alexander :nalexander [he/him]]

mcomella: you know the most about Intent handling.  Can you investigate or re-direct?

Comment 2

[Sebastian Kaspari (:sebastian; :pocmo)]

Shouldn't the action be android.intent.action.VIEW (instead of android.activity.MAIN)?

Comment 3

[Michael Comella (:mcomella) [NI reported issues only: ex-Mozilla]]

(In reply to Sebastian Kaspari (:sebastian) from comment #2)
> Shouldn't the action be android.intent.action.VIEW (instead of
> android.activity.MAIN)?

Sounds feasible.

Margaret, I'd like to focus on telemetry – do you have time to look into this further?

Comment 4

[:Margaret Leibovic]

Nick, did you just accidentally specify action.MAIN instead of action.VIEW?

Or is there something more we need to look into here?

Comment 5

[Nick Alexander :nalexander [he/him]]

(In reply to :Margaret Leibovic from comment #4)
> Nick, did you just accidentally specify action.MAIN instead of action.VIEW?

Accidentally, in the sense that I don't really care -- I've done this for years, following http://wrla.ch/blog/2012/05/launching-random-web-browsers-on-android-via-adb/?utm_source=rss&utm_medium=rss&utm_campaign=launching-random-web-browsers-on-android-via-adb

> Or is there something more we need to look into here?

adb shell am start -a android.activity.VIEW -n org.mozilla.fennec_nalexander/org.mozilla.gecko.BrowserApp -d 'http://people.mozilla.org/~ewong2/push-notification-test/'
Starting: Intent { act=android.activity.VIEW dat=http://people.mozilla.org/~ewong2/push-notification-test/ cmp=org.mozilla.fennec_nalexander/org.mozilla.gecko.BrowserApp }
Warning: Activity not started, its current task has been brought to the front

and

adb shell am start -a android.activity.VIEW -n org.mozilla.fennec_nalexander/.App -d 'http://people.mozilla.org/~ewong2/push-notification-test/'

both behave the same for me: Fennec focused, no tab opened.  (Including when there are no tabs open -- i.e., we're not just focusing an existing tab in some instances.)

Comment 6

[:Margaret Leibovic]

Is this a recent regression for you? This was working for me just earlier today.

Comment 7

[Nick Alexander :nalexander [he/him]]

(In reply to :Margaret Leibovic from comment #6)
> Is this a recent regression for you? This was working for me just earlier
> today.

No, months, across several devices (I think).  What was your exact command line?

Comment 8

[Nick Alexander :nalexander [he/him]]

I haven't a clue why my old approach stopped working, but it appears the following works.  Observe the different action, and the less specific package name.

adb shell am start -a android.intent.action.VIEW -d 'http://people.mozilla.org/~ewong2/push-notification-test/' org.mozilla.fennec_nalexander

Comment 9

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=f1fbf9e08300

Comment 10

[Nick Alexander :nalexander [he/him]]

I actually think there are some issues in the existing code, and I want to change how this works as part of Bug 1242213, so I'm re-opening it.

Comment 11

[Nick Alexander :nalexander [he/him]]

Attachment: MozReview Request: Bug 1237755 - Part 1: Use android.intent.action.{MAIN,VIEW} in |mach run|. r?gbrown

The motivation is to remove the hard-coded ".App" that we were using:
there's no reason to not use the intent filters in the same manner
that Android itself will.  Along the way, I replaced
"android.activity.MAIN" because I don't really understand how that
worked at any point: I can't find any documentation referencing it!

Review commit: https://reviewboard.mozilla.org/r/32217/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/32217/

Comment 12

[Nick Alexander :nalexander [he/him]]

Attachment: MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

Review commit: https://reviewboard.mozilla.org/r/32219/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/32219/

Comment 13

[Nick Alexander :nalexander [he/him]]

Attachment: MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

Review commit: https://reviewboard.mozilla.org/r/32221/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/32221/

Comment 14

[Nick Alexander :nalexander [he/him]]

gbrown: this is motivated by Bug 1242213, which gets rid of .App entirely.

There's something very fishy with this code, since there appears to be a near-complete duplicate of the launching code in mozdevice.  In any case, could you take a look, and could you suggest any places automation might be launching with an explicit .App in non-tree code?  (I hope there are none.)

Comment 15

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=77a1e3c11378

Comment 16

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=3b05be695e01

Comment 17

[Geoff Brown [:gbrown]]

(In reply to Nick Alexander :nalexander from comment #14)
> could you take a look, and could you suggest any places automation might be
> launching with an explicit .App in non-tree code?  (I hope there are none.)

There's probably an .App or two in autophone?

Comment 18

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711502 [details]
MozReview Request: Bug 1237755 - Part 1: Use android.intent.action.{MAIN,VIEW} in |mach run|. r?gbrown

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32217/diff/1-2/

Comment 19

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32219/diff/1-2/

Comment 20

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711504 [details]
MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32221/diff/1-2/

Comment 21

[Geoff Brown [:gbrown]]

Comment on attachment 8711502 [details]
MozReview Request: Bug 1237755 - Part 1: Use android.intent.action.{MAIN,VIEW} in |mach run|. r?gbrown

https://reviewboard.mozilla.org/r/32217/#review28949

Works great for me.

Comment 22

[Bob Clary [:bc] (inactive)]

Autophone uses a forked copy of adb.py, adb_android.py and as such will be affected when I merge the two.

Note: I noticed two trailing whitespace issues in a quick look at the adb_android.py patch.

Question: Will these changes work with older versions of Fennec? If not, then we will need a fallback.

Comment 23

[Nick Alexander :nalexander [he/him]]

(In reply to Bob Clary [:bc:] from comment #22)
> Autophone uses a forked copy of adb.py, adb_android.py and as such will be
> affected when I merge the two.
> 
> Note: I noticed two trailing whitespace issues in a quick look at the
> adb_android.py patch.

Oh!  Thanks!  I will take a look at clean whitespace.

> Question: Will these changes work with older versions of Fennec? If not,
> then we will need a fallback.

This is a fine question.  It's my belief that this will work without change with older versions of Fennec, since the mechanism is strictly more general.  If there's a way to verify this short of landing, I'd appreciate to know.

Comment 24

[Geoff Brown [:gbrown]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

https://reviewboard.mozilla.org/r/32219/#review28955

::: testing/mozbase/mozdevice/mozdevice/adb_android.py:347
(Diff revision 2)
> +            

Extra whitespace

::: testing/mozbase/mozdevice/mozdevice/droid.py:67
(Diff revision 2)
> +            

Extra whitespace

Comment 25

[Geoff Brown [:gbrown]]

Comment on attachment 8711504 [details]
MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

https://reviewboard.mozilla.org/r/32221/#review28957

You can run these tests right from the directory with (hopefully!) no setup: python droidsut_launch.py

(It's not passing for me.)

::: testing/mozbase/mozdevice/tests/droidsut_launch.py:14
(Diff revision 2)
> -                                         "org.mozilla.fennec/.App -a "
> +                                         "-a android.intent.action.MAIN"

missing ' ' after .MAIN here (string concatenation)

::: testing/mozbase/mozdevice/tests/droidsut_launch.py:27
(Diff revision 2)
> -                                         "org.mozilla.fennec/.App -a "
> +                                         "-a android.intent.action.MAIN"

another missing ' '

Comment 26

[Bob Clary [:bc] (inactive)]

fyi, I'm running a test on autophone with the changes to adb_android.py using Nexus One CyanogenMod 2.3.6, Samsung Galaxy S III Android 4.0.4 and Nexus 6P Android 6.0.1. I'll let you know how it goes.

Comment 27

[Bob Clary [:bc] (inactive)]

Note in part 2 for adb_android.py:

        if activityName:
            acmd.extend(["-n", "%s/%s" % (appName, activityName)])
        else:
            acmd.extend([appName])

should use activity_name instead of activityName

Comment 28

[Geoff Brown [:gbrown]]

(In reply to Nick Alexander :nalexander from comment #14)
> There's something very fishy with this code, since there appears to be a
> near-complete duplicate of the launching code in mozdevice.

There is lots of overlap between devicemanagerADB.py (older, used in traditional automated unit tests) and {adb.py, adb_android.py} (newer, out-of-tree copy of these used in autophone): they serve similar purposes in different projects. It looks to me like you (almost, modulo :bc's comments!) made the right changes in both places. Thanks!

Comment 29

[Bob Clary [:bc] (inactive)]

ditto appName vs app_name.

Even with this change, it doesn't appear that this launches current mozilla-inbound fennec builds in autophone.

Comment 30

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32219/diff/2-3/

Comment 31

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711504 [details]
MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32221/diff/2-3/

Comment 32

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=7c3c225fe489

Comment 33

[Geoff Brown [:gbrown]]

Comment on attachment 8711504 [details]
MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

https://reviewboard.mozilla.org/r/32221/#review29065

droidsut_launch.py passes for me now - thanks.

Comment 34

[Geoff Brown [:gbrown]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

https://reviewboard.mozilla.org/r/32219/#review29069

I am troubled by https://bugzilla.mozilla.org/show_bug.cgi?id=1237755#c29. We need to make sure that your adb_android.py changes do not break autophone. I'll investigate tomorrow, unless :bc beats me to it! ;)

::: testing/mozbase/mozdevice/mozdevice/adb_android.py:356
(Diff revision 3)
> +            acmd.extend(["-n", "%s/%s" % (appName, activity_name)])

appName -> app_name

::: testing/mozbase/mozdevice/mozdevice/adb_android.py:358
(Diff revision 3)
> +            acmd.extend([appName])

appName -> app_name

Comment 35

[Bob Clary [:bc] (inactive)]

Attachment: foo.py

This is a rough test showing the problem. This fails to load urls from the local file system but does load remote urls such as http://cnn.com/. It uses the serial number for my gs3 and paths specific to autophone but should be easily modifiable.

Comment 36

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=22e4080a638a

Comment 37

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711502 [details]
MozReview Request: Bug 1237755 - Part 1: Use android.intent.action.{MAIN,VIEW} in |mach run|. r?gbrown

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32217/diff/2-3/

Comment 38

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32219/diff/3-4/

Comment 39

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8711504 [details]
MozReview Request: Bug 1237755 - Part 3: Update mozdevice tests. r?gbrown

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/32221/diff/3-4/

Comment 40

[Bob Clary [:bc] (inactive)]

Attachment: bug-1237755-autophone.patch

jmaher, this works well with nalexander's changes for s1s2, but fails for talos. I also tried the original -tp file:/sdcard/... in addition to -tp file:///sdcard/... but neither works. Thoughts?

Comment 41

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=7f42947c07a3

Comment 42

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=e4e419090d5d

Comment 43

[Nick Alexander :nalexander [he/him]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=8bfca15075f3

Comment 44

[Nick Alexander :nalexander [he/him]]

gbrown: bc: here's where this is at.  Lots of try pushes have finished all green M or R.  The try push at

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c2f31253bcfa

has a green tpn run, but trychooser lies about what the syntax for "the other talos test" is and I'm done fighting this.  (I already discovered that -t all is broken.)  Let's assume the other talos test doesn't do wildly different things, since we're confident it's all the same pageloader extension, which needs no changes.

To the best of my knowledge, the patch is working in bc's autophone tests after removing an explicit "intent=...VIEW" line.

So, back to you for final review.

Comment 45

[Bob Clary [:bc] (inactive)]

Attachment: bug-1237755-adjust-mimetype-for-file-urls.patch

This fixes loading the about:fennec page in smoketest:

https://treeherder.allizom.org/#/jobs?repo=mozilla-inbound&revision=fd3b8111af34&exclusion_profile=false&filter-tier=1&filter-tier=2&filter-tier=3&filter-searchStr=autophone

Comment 46

[Bob Clary [:bc] (inactive)]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

https://reviewboard.mozilla.org/r/32219/#review28951

::: testing/mozbase/mozdevice/mozdevice/adb_android.py:347
(Diff revision 2)
> +            

whitespace

::: testing/mozbase/mozdevice/mozdevice/droid.py:67
(Diff revision 2)
> +            

whitepace

::: testing/mozbase/mozdevice/mozdevice/adb_android.py:357
(Diff revision 4)
> +            acmd.extend(["-t", "text/html"])

Since we only care about file: urls specifically here, we should limit this mimetype override to that case. Otherwise this will prevent loading of about: type urls. See the attached patch which makes this work for about:fennec urls.

Comment 47

[Bob Clary [:bc] (inactive)]

Attachment: bug-1237755-no-intent-v1.patch

Comment 48

[Bob Clary [:bc] (inactive)]

Attachment: bug-1237755-explicit-file-scheme-v1.patch

Comment 49

[Nick Alexander :nalexander [he/him]]

(In reply to Bob Clary [:bc:] from comment #46)
> Comment on attachment 8711503 [details]
> MozReview Request: Bug 1237755 - Part 2: Use
> android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc
> 
> https://reviewboard.mozilla.org/r/32219/#review28951
> 
> ::: testing/mozbase/mozdevice/mozdevice/adb_android.py:347
> (Diff revision 2)
> > +            
> 
> whitespace
> 
> ::: testing/mozbase/mozdevice/mozdevice/droid.py:67
> (Diff revision 2)
> > +            
> 
> whitepace
> 
> ::: testing/mozbase/mozdevice/mozdevice/adb_android.py:357
> (Diff revision 4)
> > +            acmd.extend(["-t", "text/html"])
> 
> Since we only care about file: urls specifically here, we should limit this
> mimetype override to that case. Otherwise this will prevent loading of
> about: type urls. See the attached patch which makes this work for
> about:fennec urls.

This is fine by me.  To be clear, we're pushing http/https/about/javascript URLs to https://dxr.mozilla.org/mozilla-central/rev/aa90f482e16db77cdb7dea84564ea1cbd8f7f6b3/mobile/android/base/AndroidManifest.xml.in#251 and file URLs to https://dxr.mozilla.org/mozilla-central/rev/aa90f482e16db77cdb7dea84564ea1cbd8f7f6b3/mobile/android/base/AndroidManifest.xml.in#258.  The second filter has a mimeType, hence needs the -t.

Comment 50

[Geoff Brown [:gbrown]]

Comment on attachment 8711503 [details]
MozReview Request: Bug 1237755 - Part 2: Use android.intent.action.{MAIN,VIEW} in mozdevice. r?gbrown,bc

https://reviewboard.mozilla.org/r/32219/#review29699

Comment 51

[Joel Maher ( :jmaher ) (UTC -8)]

Comment on attachment 8713271 [details] [diff] [review]
bug-1237755-autophone.patch

Review of attachment 8713271 [details] [diff] [review]:
-----------------------------------------------------------------

we chatted in irc, in the past there have been issues with -tp file:<path> to pageloader, but those issues could have been os version specific.  If this works, I don't see any other big issues with it.

Comment 52

[Bob Clary [:bc] (inactive)]

I ran some tests with this patch using one of my older Nexus One running CyanogenMod and found that it doesn't launch the browser but instead launches a chooser between Browser or Nightly with the option to make it the default action.

adb -s HT09AP802284 wait-for-device shell am start -a android.intent.action.VIEW --es env3 MOZ_DISABLE_NONLOCAL_CONNECTIONS=1 --es env2 NO_EM_RESTART=1 --es env1 MOZ_CRASHREPORTER_NO_REPORT=1 --es env0 MOZ_CRASHREPORTER=1 --es env4 MOZ_CRASHREPORTER_SHUTDOWN=1 -d file:///sdcard/tests/autophone/S1S2Test/files/s1s2/blank.html -t text/html org.mozilla.fennec
Starting: Intent { act=android.intent.action.VIEW dat=file:///sdcard/tests/autophone/S1S2Test/files/s1s2/blank.html typ=text/html }

Gives a choice between HTMLViewer and Nightly while

adb -s HT09AP802284 wait-for-device shell am start -a android.intent.action.VIEW --es env3 MOZ_DISABLE_NONLOCAL_CONNECTIONS=1 --es env2 NO_EM_RESTART=1 --es env1 MOZ_CRASHREPORTER_NO_REPORT=1 --es env0 MOZ_CRASHREPORTER=1 --es env4 MOZ_CRASHREPORTER_SHUTDOWN=1 -d http://cnn.com -t text/html org.mozilla.fennec
Starting: Intent { act=android.intent.action.VIEW dat=http://cnn.com typ=text/html }

gives a choice between Browser and Nightly.

I'll run additional tests with other Nexus One devices I have to make sure it is a common behavior.

nalexander: Does the new am start syntax work with Android 2.3?

Comment 53

[Nick Alexander :nalexander [he/him]]

(In reply to Bob Clary [:bc:] from comment #52)
> I ran some tests with this patch using one of my older Nexus One running
> CyanogenMod and found that it doesn't launch the browser but instead
> launches a chooser between Browser or Nightly with the option to make it the
> default action.
> 
> adb -s HT09AP802284 wait-for-device shell am start -a
> android.intent.action.VIEW --es env3 MOZ_DISABLE_NONLOCAL_CONNECTIONS=1 --es
> env2 NO_EM_RESTART=1 --es env1 MOZ_CRASHREPORTER_NO_REPORT=1 --es env0
> MOZ_CRASHREPORTER=1 --es env4 MOZ_CRASHREPORTER_SHUTDOWN=1 -d
> file:///sdcard/tests/autophone/S1S2Test/files/s1s2/blank.html -t text/html
> org.mozilla.fennec
> Starting: Intent { act=android.intent.action.VIEW
> dat=file:///sdcard/tests/autophone/S1S2Test/files/s1s2/blank.html
> typ=text/html }
> 
> Gives a choice between HTMLViewer and Nightly while
> 
> adb -s HT09AP802284 wait-for-device shell am start -a
> android.intent.action.VIEW --es env3 MOZ_DISABLE_NONLOCAL_CONNECTIONS=1 --es
> env2 NO_EM_RESTART=1 --es env1 MOZ_CRASHREPORTER_NO_REPORT=1 --es env0
> MOZ_CRASHREPORTER=1 --es env4 MOZ_CRASHREPORTER_SHUTDOWN=1 -d http://cnn.com
> -t text/html org.mozilla.fennec
> Starting: Intent { act=android.intent.action.VIEW dat=http://cnn.com
> typ=text/html }
> 
> gives a choice between Browser and Nightly.
> 
> I'll run additional tests with other Nexus One devices I have to make sure
> it is a common behavior.
> 
> nalexander: Does the new am start syntax work with Android 2.3?

Apparently not: 2.3.7 doesn't accept bare package names: http://androidxref.com/2.3.7/xref/frameworks/base/core/java/android/content/ComponentName.java#159.

Good grief.  OK, I'll just s/.App/org.mozilla.gecko.BrowserApp/ and give up trying to make this better.  Thanks for testing, bc!

Comment 54

[Bob Clary [:bc] (inactive)]

We change launch_fennec to make activity_name an optional keyword argument. It would be up to the caller to decide if to pass it and we could still use your new syntax for Android 4.0.3+.

After all, 2.3 support is going to die sometime in the relative near future.

One thing though, is with your changes, *one* of my Nexus 6Ps fails the smoke test. Using the original approach it does not fail. It appears to have a permissions problem with the profile on the sdcard. The other 5 6Ps I have here locally don't have that problem and I don't understand what is different between the two sets of devices. I tried working around that by using /data/local instead of /sdcard and it resolved the smoketest issue, but caused the unit tests to fail. :-( I was planning on investigating that next.

If s/.App/org.mozilla.gecko.BrowserApp/ gives you want you need, lets go with that. I'll test your patch when you have it ready. We can revisit this when we drop support for 2.3 on release.

Thanks for your patience.

Comment 55

[Bob Clary [:bc] (inactive)]

Attachment: launch_application.patch

I tested this in autophone along with the previous autophone patches to use file: urls on the smoketest, talos, s1s2test and webappstartup tests on Android 2.3, Android 4.0.3 and Android 6.0.1 and got full green on all including the one 6P which had the orange on the smoketest:

https://treeherder.allizom.org/#/jobs?repo=mozilla-inbound&revision=62d300c9c733&exclusion_profile=false&filter-tier=1&filter-tier=2&filter-tier=3&filter-searchStr=autophone

nalexander: You'll need to do a patch for mozdevice's adb_android.py and probably should do a try run on the unit tests. I think this will be sufficient:

try: -b o -p android-api-9,android-api-15 -u plain-reftest-1,jsreftest-1,mochitest-1 -t remote-tsvg,remote-tp4m_nochrome

Comment 56

[Nick Alexander :nalexander [he/him]]

Comment on attachment 8714991 [details] [diff] [review]
launch_application.patch

Review of attachment 8714991 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm.  Ta.  This will work with everything right now -- you can land independent of anything else in the tree.  (And it is the "new normal", so we're unlikely to need to roll back.)

Comment 57

[Bob Clary [:bc] (inactive)]

I commit the following to Autophone, but haven't made any changes to mozbase/mozdevice or any test runners.

https://github.com/mozilla/autophone/commit/656627556c82e76fb6e16374b7ea31775734e17f
Bug 1237755 - Autophone - adb_android.py - use activity name org.mozilla.gecko.BrowserApp when launching Fennec, r=nalexander.

https://github.com/mozilla/autophone/commit/af4126e14578f0ce52e9324234a7990b7fdd2279
Bug 1237755 - Autophone - use explicit file: scheme for local paths, r=gbrown

I'll deploy these to autophone production later.

nalexander: I'm not sure what other changes you will still be making to mozbase/mozdevice as part of this bug, but you'll need to work up an equivalent patch for https://github.com/mozilla/autophone/commit/656627556c82e76fb6e16374b7ea31775734e17f to apply to mozdevice's version of adb_android.py.

Comment 58

[Nick Alexander :nalexander [he/him]]

Whatever is going on here is no longer relevant.