1049790 - [B2G][FTU][Date & Time] Changes to Continent and City are not displayed properly.

Status: VERIFIED FIXED

(Firefox OS Graveyard :: Gaia::First Time Experience, defect)

Description

[Martin Shuman [:Marty]]

Description:
In the 'Date & Time' screen of the FTU, the user is able to Set the Continent and City.  Changes made to these settings don't always update the display, requiring the user to switch back and forth several times for the change to show properly.


Repro Steps:
1) Update a Flame to 20140806000200
2) Progress through the FTU to the 'Date & Time' screen
3) Change the Continent and City settings.


Actual:
User choices are stored, but not displayed.


Expected:
User choices are displayed properly.

Environmental Variables:
Device: Flame 2.0 (319MB)
Build ID: 20140806000200
Gaia: 5ba22d458fdb63bd72c59de53c701d0efe35c1e2
Gecko: 6fbc60a80c6d
Version: 32.0 (2.0)
Firmware Version: v122
User Agent: Mozilla/5.0 (Mobile; rv:32.0) Gecko/32.0 Firefox/32.0


Keywords: FTU, FTE, Date, Time, Continent, City, Update, Display


Repro frequency: ~50%
Link to failed test case: https://moztrap.mozilla.org/manage/case/6119/
See attached: video clip

Comment 1

[Martin Shuman [:Marty]]

Issue does not seem to occur on Flame 2.0 (512MB)
Continent and City display updates properly.

Environmental Variables:
Device: Flame 2.0 (512MB)
Build ID: 20140806000200
Gaia: 5ba22d458fdb63bd72c59de53c701d0efe35c1e2
Gecko: 6fbc60a80c6d
Version: 32.0 (2.0)
Firmware Version: v122
User Agent: Mozilla/5.0 (Mobile; rv:32.0) Gecko/32.0 Firefox/32.0

-------------------------------------------------------------------

I am unable to see if Flame 2.1 is affected due to bug 1049149.

Comment 2

[Peter Bylenga [:PBylenga]]

[Blocking Requested - why for this release]:
This was filed as a follow up to bug 1042388, it's a regression of a core feature and a smoketest blocker.

Comment 3

[Gregor Wagner [:gwagner]]

Fernando, any idea whats going on here?

Comment 4

[Jason Smith [:jsmith]]

Can someone try reproducing this with public Flame builds:

ftp://ftp.mozilla.org/pub/b2g/nightly/latest-mozilla-b2g32_v2_0-flame/

Comment 5

[Fernando Campo (:fcampo)]

Well, one possibility is that my patch for bug 1042388 wasn't working after all, and that we hit what Tim describes in bug 1042388 comment 15:
> Also, this bug cannot be reproduced once I do another |adb reboot| in attempt to update ftu package
> again, so I highly doubt this bug should remain block. But QA would probably disagree because this
> apparently blocks the very important smoketest from continue.
If the patch was applied after flashing (with full startup of the phone), and then a reset-gaia was made, that could have led to a false positive.

I'm still unable to repro with TEF builds, but I'll try with the public one (that I actually didn't know existed!)

Comment 6

[Jason Smith [:jsmith]]

(In reply to Jason Smith [:jsmith] from comment #4)
> Can someone try reproducing this with public Flame builds:
> 
> ftp://ftp.mozilla.org/pub/b2g/nightly/latest-mozilla-b2g32_v2_0-flame/

Btw - you'll want download the b2g-32.0.en-US.android-arm.tar.gz & gaia.zip, unzip the files in the same root directory, and download https://raw.githubusercontent.com/nhirata/fullflash/master/flash_Gg.sh. That should result in the following setup:

b2g/
gaia/
flash_Gg.sh

Then, run ./flash_Gg.sh.

Comment 7

[Fernando Campo (:fcampo)]

So bad news, I am able to repro with the public build.
Better than bad news (still not good) is that I can try to work on it now.

Thanks for the link and explanation Jason

Comment 8

[Joshua Mitchell (Inactive)]

(In reply to Jason Smith [:jsmith] from comment #6)
> (In reply to Jason Smith [:jsmith] from comment #4)
> > Can someone try reproducing this with public Flame builds:

This issue DOES reproduce in this 2.0 Public build

Actual Results - not all changes are updated on the map (properly displayed)

additionally - not all the changes are 'stored' correctly - meaning that sometimes when I select a different option (ex: Asia) it will  turn blue and get the blue checkmark, but when I hit OK the change is not reflected on the Date & Time page. Returning to the specific list (country or city) the new choice will still show highlighted blue with the checkmark and it can take several times going back and forth before the field is update. However; this only occurred during the first FTU encounter and did not repro later by launching the FTU from the developer menu

Device: Flame 2.0
Build ID: 20140807000201
Gaia: 8cc28fd31905a0ea2b2e15d13e80a0eab2feb1ba
Gecko: 25980b5120b0
Version: 32.0 (2.0)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:32.0) Gecko/32.0 Firefox/32.0

Notes: Sim in, no wi-fi connection, no cellular data turned on

Comment 10

[Ralph Giles (:rillian)]

I've also hit this problem with recent vendor builds for the flatfish tablet.

Comment 11

[Fernando Campo (:fcampo)]

So I've been working on this for the last few days, tried more things during the weekend, and I'm out of ideas, still nothing works out, so some help would be most appreciated. To sum up:
- only happens on 273Mb flame. With more memory or other devices I'm unable to repro, no matter the build (didn't try on flatfish)
- only happens on fresh flash. The moment that we make a reset-gaia (even putting in the same gaia commit that it's already flashed, the problem disappears)
- only on MOZ builds, not on TEF side ones. I don't know if it might be related to the build process.

Here's what I've already tried:
- eliminated the duplicated call to tz_select - patch for bug bug 1042388 - in case the events were catched by the wrong object.
- trigger the changes before changing the timezone setting, in case they were not launched.
- created the 'moztimechange' listener before setting the timezone setting, in case it was too quick.

None of those solved the problem, and as I said, I'm out of ideas :(
If anybody have any clue, please jump in.

Comment 12

[Fernando Campo (:fcampo)]

I'm reassigning myself to let room to anyone with more ideas, although I'll keep trying things.
ni? gregor so he can proxy my petition for help

Comment 13

[Fernando Campo (:fcampo)]

Jeez, I need a break...I meant de-assigning

Comment 14

[Jason Smith [:jsmith]]

Marking qawanted here to see if this happens on 319 MB Flame on 1.4. If it doesn't, let's try getting a window here.

Comment 15

[Gregor Wagner [:gwagner]]

Aus, can you give it a try?

Comment 16

[Aus Lacroix [:aus]]

I'll take a stab at this. What's the best way to repro? Looks like a fresh flash using Naoki's script is all it takes? Or do we need 273M memory profile as well?

Comment 17

[Jason Smith [:jsmith]]

(In reply to Ghislain Aus Lacroix [:aus] from comment #16)
> I'll take a stab at this. What's the best way to repro? Looks like a fresh
> flash using Naoki's script is all it takes? Or do we need 273M memory
> profile as well?

My understanding is that you need to flash a build from pvtbuilds (https://pvtbuilds.mozilla.org/pvt/mozilla.org/b2gotoro/nightly/mozilla-b2g32_v2_0-flame/latest/) and try to reproduce this bug under a 319 MB Flame setting.

Comment 18

[Aus Lacroix [:aus]]

Occurs as reported with 319M memory profile with 2.0 and 2.1.

Comment 19

[Duane Dixon [:ddixon]]

Following results checked on device with 319 MB memory. 

DOES occur in Flame 1.4 with v122 base image.

Device: Flame 1.4
Build ID: 20140811081112
Gaia: d47fecdcaa4aefb754561114edd22fb23a92b98d
Gecko: fbee340d0830
Version: 30.0 (1.4)
Firmware Version: v122
User Agent: Mozilla/5.0 (Mobile; rv:30.0) Gecko/30.0 Firefox/30.0


DOES occur in Flame 1.4 with v123 base image. 

Device: Flame 1.4
Build ID: 20140811081112
Gaia: d47fecdcaa4aefb754561114edd22fb23a92b98d
Gecko: fbee340d0830
Version: 30.0 (1.4)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:30.0) Gecko/30.0 Firefox/30.0

Comment 20

[Jason Smith [:jsmith]]

Duane - Can you also check the base image as well?

Comment 21

[Duane Dixon [:ddixon]]

DOES NOT occur on v122, v123 base images with 319 MB memory.

Device: Flame 1.3
Build ID: 20140627162151
Gaia: 5c43d012a9351e8aaeacda0b87b0886b7f67d33d
Gecko: e181a36ebafaa24e5390db9f597313406edfc794
Version: 28.0 (1.3)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:28.0) Gecko/28.0 Firefox/28.0

Device: Flame 1.3
Build ID: 20140616171114
Gaia: e1b7152715072d27e0880cdc6b637f82fa42bf4e
Gecko: e181a36ebafaa24e5390db9f597313406edfc794
Version: 28.0 (1.3)
Firmware Version: v122
User Agent: Mozilla/5.0 (Mobile; rv:28.0) Gecko/28.0 Firefox/28.0

Comment 22

[Aus Lacroix [:aus]]

After investigating into this yesterday and this morning it would appear that the moztimechange event is delivered late, somehow. I only get the event the time zone is changed again. There's something weird with the event queue which is causing the event to get stuck until a new event of the same type is put into it.

Comment 23

[Aus Lacroix [:aus]]

Hah! Awesome, well, that was easier than I thought. Pull Request incoming which will cover both 2.1 and 2.0.

Comment 24

[Aus Lacroix [:aus]]

Attachment: Pull Request (v2.0 or less) - Use one listener for moztimechange

No one other than you has reviewed code for this in over a year so, lucky you! :)

Comment 25

[Aus Lacroix [:aus]]

Attachment: Pull Request (master) - Use one listener for moztimechange

Comment 27

[Fernando Campo (:fcampo)]

Comment on attachment 8471973 [details] [review]
Pull Request (master) - Use one listener for moztimechange

Code is good, and it seems to work (but again, everything seems to work when making reset-gaia :D)

Hopefully this should work when flashing a clean build, thanks.

Comment 28

[Fernando Campo (:fcampo)]

Comment on attachment 8471940 [details] [review]
Pull Request (v2.0 or less) - Use one listener for moztimechange

Patch for 2.0 is good to go too.

Comment 29

[Aus Lacroix [:aus]]

Commit (v2.0): https://github.com/mozilla-b2g/gaia/commit/408e24a54fb13217428fa9387ca01044a1ea3da4

Commit (master): https://github.com/mozilla-b2g/gaia/commit/c9127a4a9b41f99675b0ad746832bf1f511493d9

Comment 30

[Aus Lacroix [:aus]]

My change for 2.0 doesn't apply cleanly to 1.4 so I'm creating yet another patch for that branch.

Comment 31

[Aus Lacroix [:aus]]

Attachment: Pull Request (v1.4) - Use one listener for moztimechange

Too big of a difference in line numbers prevented the 2.0 PR from applying on 1.4 so I fixed the merged conflicts by hand and created this PR.

Comment 32

[Aus Lacroix [:aus]]

Commit (v1.4): https://github.com/mozilla-b2g/gaia/commit/ab9a57d204c8c06cccc70cb7190da789fe17a3cb

For my flatfish friends. :)

Comment 33

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

This broke gaia-linter on 2.0: https://tbpl.mozilla.org/php/getParsedLog.php?id=45966627&tree=Mozilla-B2g32-v2.0
13:27:06     INFO - Running command: ['make', 'lint'] in /builds/slave/test/gaia
13:27:06     INFO - Copy/paste: make lint
13:27:06     INFO - Calling ['make', 'lint'] with output_timeout 330
13:27:06     INFO -  NO_XFAIL=1 make -k gjslint hint
13:27:07     INFO -  make[1]: Entering directory `/builds/slave/test/gaia'
13:27:07     INFO -  # gjslint --disable 210,217,220,225 replaces --nojsdoc because it's broken in closure-linter 2.3.10
13:27:07     INFO -  # http://code.google.com/p/closure-linter/issues/detail?id=64
13:27:07     INFO -  Running gjslint...
13:28:17     INFO -  542 files checked, no errors found.
13:28:17     INFO -  Note: gjslint only checked the files that are xfailed for jshint.
13:28:17     INFO -  Running jshint...
13:30:54    ERROR - TEST-UNEXPECTED-FAIL | shared/js/tz_select.js | line 208, col 8, Missing semicolon.
13:30:54     INFO -  shared/js/tz_select.js: line 208, col 8, Missing semicolon. (ERROR)
13:30:54    ERROR - TEST-UNEXPECTED-FAIL | shared/js/tz_select.js | line 231, col 9, Missing semicolon.
13:30:54     INFO -  shared/js/tz_select.js: line 231, col 9, Missing semicolon. (ERROR)
13:30:54     INFO -  2 errors (13275 xfailed)
13:30:54     INFO -  Please consult https://github.com/mozilla-b2g/gaia/tree/master/build/jshint/README.md to get some information about how to fix jshint issues.
13:30:54     INFO - TEST-UNEXPECTED-FAIL | make lint | make[1]: *** [hint] Error 1
13:30:54     INFO -  make[1]: *** [hint] Error 1
13:30:54     INFO -  make[1]: Leaving directory `/builds/slave/test/gaia'
13:30:54     INFO - TEST-UNEXPECTED-FAIL | make lint | make: *** [lint] Error 2
13:30:54     INFO -  make: *** [lint] Error 2
13:30:54    ERROR - Return code: 2
13:30:54     INFO - TinderboxPrint: gaia-lint: <em class="testfail">4 errors</em>
13:30:54    ERROR - Tests exited with return code 2: harness failures
13:30:54    ERROR - # TBPL FAILURE #

Comment 34

[Aus Lacroix [:aus]]

(In reply to Wes Kocher (:KWierso) from comment #33)
> This broke gaia-linter on 2.0:
> https://tbpl.mozilla.org/php/getParsedLog.php?id=45966627&tree=Mozilla-B2g32-
> v2.0
> 13:27:06     INFO - Running command: ['make', 'lint'] in
> /builds/slave/test/gaia
> 13:27:06     INFO - Copy/paste: make lint
> 13:27:06     INFO - Calling ['make', 'lint'] with output_timeout 330
> 13:27:06     INFO -  NO_XFAIL=1 make -k gjslint hint
> 13:27:07     INFO -  make[1]: Entering directory `/builds/slave/test/gaia'
> 13:27:07     INFO -  # gjslint --disable 210,217,220,225 replaces --nojsdoc
> because it's broken in closure-linter 2.3.10
> 13:27:07     INFO -  #
> http://code.google.com/p/closure-linter/issues/detail?id=64
> 13:27:07     INFO -  Running gjslint...
> 13:28:17     INFO -  542 files checked, no errors found.
> 13:28:17     INFO -  Note: gjslint only checked the files that are xfailed
> for jshint.
> 13:28:17     INFO -  Running jshint...
> 13:30:54    ERROR - TEST-UNEXPECTED-FAIL | shared/js/tz_select.js | line
> 208, col 8, Missing semicolon.
> 13:30:54     INFO -  shared/js/tz_select.js: line 208, col 8, Missing
> semicolon. (ERROR)
> 13:30:54    ERROR - TEST-UNEXPECTED-FAIL | shared/js/tz_select.js | line
> 231, col 9, Missing semicolon.
> 13:30:54     INFO -  shared/js/tz_select.js: line 231, col 9, Missing
> semicolon. (ERROR)
> 13:30:54     INFO -  2 errors (13275 xfailed)
> 13:30:54     INFO -  Please consult
> https://github.com/mozilla-b2g/gaia/tree/master/build/jshint/README.md to
> get some information about how to fix jshint issues.
> 13:30:54     INFO - TEST-UNEXPECTED-FAIL | make lint | make[1]: *** [hint]
> Error 1
> 13:30:54     INFO -  make[1]: *** [hint] Error 1
> 13:30:54     INFO -  make[1]: Leaving directory `/builds/slave/test/gaia'
> 13:30:54     INFO - TEST-UNEXPECTED-FAIL | make lint | make: *** [lint]
> Error 2
> 13:30:54     INFO -  make: *** [lint] Error 2
> 13:30:54    ERROR - Return code: 2
> 13:30:54     INFO - TinderboxPrint: gaia-lint: <em class="testfail">4
> errors</em>
> 13:30:54    ERROR - Tests exited with return code 2: harness failures
> 13:30:54    ERROR - # TBPL FAILURE #

I already pushed a specific fix for this issue. It should recover when the next build cycles (iirc).

Comment 35

[Tony Chung [:tchung]]

marty, please verify fix on 2.0 and master.  thanks.

Comment 36

[Martin Shuman [:Marty]]

Verified Fixed. Issue no longer occurs on Flame 1.4, Flame 2.0, or Flame 2.1 at either 512MB or 319MB.
Continent and City display updates properly.

Environmental Variables:
Device: Flame 1.4
Build ID: 20140815063005
Gaia: ab9a57d204c8c06cccc70cb7190da789fe17a3cb
Gecko: 3da3115ee33b
Version: 30.0 (1.4)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:30.0) Gecko/30.0 Firefox/30.0

Environmental Variables:
Device: Flame 2.0
Build ID: 20140815000200
Gaia: 295c7f50707372e5af6d8e83148d2d970076dfd6
Gecko: 879c5208084f
Version: 32.0 (2.0)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:32.0) Gecko/32.0 Firefox/32.0

Environmental Variables:
Device: Flame Master
Build ID: 20140815040204
Gaia: 9979fccc6321be72cd17370f3a20c65bc376af33
Gecko: c9f8cc9ce89c
Version: 34.0a1 (Master)
Firmware Version: v123
User Agent: Mozilla/5.0 (Mobile; rv:34.0) Gecko/34.0 Firefox/34.0

Comment 37

[Ryan VanderMeulen [:RyanVM]]

Reverted from v1.4 for causing frequent Gij failures (confirmed by 30+ green retriggers on the push prior).
v1.4: https://github.com/mozilla-b2g/gaia/commit/21bec64497dc06a7f12071d573570ba8fea598ae

https://tbpl.mozilla.org/php/getParsedLog.php?id=46069500&tree=Mozilla-B2g30-v1.4

Comment 38

[Aus Lacroix [:aus]]

Thanks for backing that out Ryan, I totally forgot to check up on it. :(

I should have a branch patch that doesn't break the world soon...

Comment 39

[Aus Lacroix [:aus]]

Attachment: Pull Request (v1.4) - Use one listener for moztimechange

Comment 40

[Aus Lacroix [:aus]]

And, we're all green!

https://tbpl.mozilla.org/?rev=9e6b60dae6f2bffffb58957b52d396f1c1a8d714&tree=Gaia-Try

Commit (v1.4): https://github.com/mozilla-b2g/gaia/commit/eb6f48b786999b833b4175cfd5bb9ef22227e65f

Fixed on v1.4.