845128 - [Twitter] [Browser] Twitter mobile website doesn't work in Browser, nor as everything.me app

Status: VERIFIED FIXED

(Firefox OS Graveyard :: General, defect)

Description

[Delphine Lebédel [:delphine - OOO]]

* Steps to reproduce:
Try opening up twitter in browser or as an everything.me app

* Expected Result:
twitter appears correctly

* Actual result:
blank screen

Tried this on v1 and v1.0.1 using latest builds

Comment 1

[Delphine Lebédel [:delphine - OOO]]

Sorry, was too quick in opening the bug, forgot lot's of stuff!
So:
This was done using Unagi Device

v1:
Gaia   539e4f57b25b839b7b3d25619c0dc53fd1d67579
BuildID 20130225070202
Version 18.0

v1.0.1:
Gaia   5691a16fff8e1403c75ed9d6f3a443b7e58198c6
BuildID 20130225070200
Version 18.0

Comment 2

[Nicolas B. Pierron [:nbp]]

(In reply to delphine from comment #0)
> * Steps to reproduce:
> Try opening up twitter in browser or as an everything.me app

https://mobile.twitter.com

Comment 4

[Jason Smith [:jsmith]]

Attachment: Logcat

Comment 5

[Jason Smith [:jsmith]]

So this doesn't look good:

03-01 10:31:17.912: E/GeckoConsole(471): [JavaScript Error: "uncaught exception: 2147500037"]

I have no clue what that is.

Comment 6

[Jason Smith [:jsmith]]

Confirmed it's a platform regression.

jsmith	fabrice: Speaking of facebook being busted, looks like twitter is busted too (bug 845128). Do you know what this means - 03-01 10:31:17.912: E/GeckoConsole(471): [JavaScript Error: "uncaught exception: 2147500037"]
	firebot	Bug https://bugzilla.mozilla.org/show_bug.cgi?id=845128 nor, --, ---, nobody, NEW, [Twitter] [Browser] Twitter mobile website doesn't work in Browser, nor as everything.me app
	fabrice	jsmith: let me look this up
	=-=	Mode #b2g +ao cjones cjones by ChanServ
	fabrice	bah, NS_ERROR_FAILURE

Comment 7

[[:fabrice] Fabrice Desré]

Very strange, since it works well as an installed app.

Comment 8

[Alex Keybl [:akeybl]]

Can we get a regression range on v1.0.1? The landing volume is so low there that it should be fairly obvious.

Comment 9

[Jason Smith [:jsmith]]

Will put an email about this soon, but only one keyword is needed if there's specificity - so regression window wanted is all we need here.

I'll get someone assigned to look into this.

Comment 10

[Chris Kreinbring [:CKreinbring]]

The everything.me part of this defect can also been seen with News > Newsorganizer.

Unagi build 20130301070202
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/69e249bf251c
Gaia: f46906e594b613571bfcc8f146d60bffd42e5a5b
Kernel Date: Dec 5

Comment 11

[Angela Hubenya]

Regression window results: 
The last time Twitter worked on the Browser and everything.me app, was on the 20130108070203 build. It did not work on any builds after that. 
 
https://pvtbuilds.mozilla.org/pub/mozilla.org/b2g/nightly/mozilla-b2g18-unagi/2013/01/2013-01-08-07-02-03/

Comment 12

[Jason Smith [:jsmith]]

I think you meant to only remove the regression window wanted keyword.

Comment 13

[Jason Smith [:jsmith]]

http://hg.mozilla.org/mozilla-central/pushloghtml?startdate=2013-01-08&enddate=2013-01-09

Comment 14

[Jason Smith [:jsmith]]

The only bug on that push log that looks like a potential cause of this regression is bug 826023. The others (bug 827806, bug 823445, bug 819852) don't really make sense.

Justin - Any ideas why twitter won't load in the browser, but will load in the app? And is bug 826023 really the cause of this regression?

Comment 15

[Axel Hecht [:Pike]]

Shouldn't this be more like http://hg.mozilla.org/releases/mozilla-b2g18/pushloghtml?startdate=2013-01-08&enddate=2013-01-09 ?

Comment 16

[Justin Lebar (not reading bugmail)]

> Any ideas why twitter won't load in the browser, but will load in the app?

Nope, sorry.

> And is bug 826023 really the cause of this regression?

Are we able to test this?  That may be better than speculating.

Comment 17

[Jason Smith [:jsmith]]

(In reply to Axel Hecht [:Pike] from comment #15)
> Shouldn't this be more like
> http://hg.mozilla.org/releases/mozilla-b2g18/pushloghtml?startdate=2013-01-
> 08&enddate=2013-01-09 ?

Yeah, that's right. Let me review that push log for comparison.

Comment 18

[Jason Smith [:jsmith]]

(In reply to Jason Smith [:jsmith] from comment #17)
> (In reply to Axel Hecht [:Pike] from comment #15)
> > Shouldn't this be more like
> > http://hg.mozilla.org/releases/mozilla-b2g18/pushloghtml?startdate=2013-01-
> > 08&enddate=2013-01-09 ?
> 
> Yeah, that's right. Let me review that push log for comparison.

Well, that didn't help too much either.

I think we're going to need to capture this NS_ERROR_FAILURE exception in a debugger and a get a stack at that point of execution. That might point us to what bug caused the regression.

Comment 19

[Angela Hubenya]

btw, I just noticed that this issue reproduces on the central build as well. In case that matters.
Tested on 2013-03-01-03-05-48

Comment 20

[Justin Lebar (not reading bugmail)]

Are we not able to bisect this issue?

Comment 21

[Jason Smith [:jsmith]]

(In reply to Justin Lebar [:jlebar] from comment #20)
> Are we not able to bisect this issue?

Oh yeah, that's a possible option.

Comment 22

[Michael Vines [:m1] [:evilmachines]]

(Feature regression, tef+ for now.  If the patch comes looking pretty risky then lets reconsider)

Comment 23

[mdaniloff]

This still reproduces on:
Unagi Build ID: 20130225070200
Kernel Date: Dec 5
Gecko: http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/3a5a27992a75
Gaia: 5691a16fff8e1403c75ed9d6f3a443b7e58198c6

Comment 24

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

The exception appears to be coming from the line https://mobile.twitter.com/:39

Can someone check what JS code is on that line?

Comment 25

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Nevermind, Blake says that that line is unrelated :(

Setting the dom.report_all_js_exceptions pref to true might help get better errors reported.

Comment 26

[Jason Smith [:jsmith]]

QA Wanted to test this with the pref specified in comment 25 set and to provide a logcat with that pref set.

Comment 27

[Andrew Overholt [:overholt]]

Attachment: logcat while loading mobile.twitter.com with pref set

I think I correctly set the pref and restarted the b2g process before attempting to load mobile.twitter.com in the browser.

$ adb shell
root@android:/ # cat /system/b2g/defaults/pref/jsdebug.js
pref("dom.report_all_js_exceptions", "true");
$ adb shell stop b2g; adb shell start b2g

Relevant snippet seems to be:

03-05 10:29:17.130 E/GeckoConsole(  624): Offline cache manifest item has Cache-control: no-store header, URL=https://mobile.twitter.com/cache/manifest
03-05 10:29:17.130 E/GeckoConsole(  624): Offline cache update error, URL=https://mobile.twitter.com/cache/manifest

Comment 28

[Jason Smith [:jsmith]]

Removing qawanted - comment 27 provides the logcat needed.

Comment 29

[Jason Smith [:jsmith]]

03-05 10:29:16.860 E/GeckoConsole(  781): [JavaScript Error: "uncaught exception: 2147500037"]
03-05 10:29:16.920 E/GeckoConsole(  781): Content JS LOG at https://mobile.twitter.com/:39 in anonymous: Revision: bcda4996e774d8152ccd4563f8c35303b0f5e5cc

Comment 30

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

I don't think the appcache warnings is what's causing the issue. And it appears that setting the pref doesn't provide any additional info. The logcat still doesn't show where the exception is actually coming from.

Comment 31

[[:fabrice] Fabrice Desré]

Attachment: patch

We are currently throwing NS_ERROR_FAILURE (the 2147500037 error code) when a page not part of an app calls mozSetMessageHandler() or mozHasPendingMessages(). This is what Twitter does when it registers itself as an activity provider. Instead of throwing, I changed that to just ignore the handler registration and return false for mozHasPendingMessages() for these pages.

Comment 32

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 721572 [details] [diff] [review]
patch

Review of attachment 721572 [details] [diff] [review]:
-----------------------------------------------------------------

Fabrice++

I marked this r+ yesterday, but apparently somehow it didn't go through :(

Comment 33

[[:fabrice] Fabrice Desré]

https://hg.mozilla.org/integration/mozilla-inbound/rev/aa858d84b69e

Comment 34

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/aa858d84b69e

Comment 35

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-b2g18/rev/8fd5bf352d45
https://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/0f8ad6685d9c

Comment 36

[Chris Kreinbring [:CKreinbring]]

As of Unagi build 20130318070202, Twitter now renders and works properly from Everything.me

Gecko: http://hg.mozilla.org/releases/mozilla-b2g18/rev/af9270e8f205
Gaia: a78ebf426840b5ef08c0cc3e437ad30aba3e2528
Kernel Date: Dec 5

Comment 37

[Chris Kreinbring [:CKreinbring]]

Also renders and works properly in the browser.