1097852 - Empty Loop landing page on Internet Explorer 10 and IE9 for link clickers

Status: VERIFIED FIXED

Description

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Very similar to bug 1073415. Most likely, something in the JS is using a feature that is not present in IE9 and IE10. We'll probably have to polyfill/patch the localization JS so that it can work in those browsers.

Comment 1

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Paul, was there anything in the IE error console when the page was blank? To get to the error console on IE, you can hit F12 on the keyboard to bring up their developer tools.

Comment 2

[Paul Silaghi, QA [:pauly]]

SEC7111: HTTPS security is compromised by http://otf.msn.com/c.gif?rid=1c1522b8bdb7499196bbffbdee5ea664&cts=1415892226355&clid=2430E160848B63B3313AE78E808B6160&di=340&mkt=en-us&pn=startpage&su=http%253A%252F%252Fwww.msn.com%252F%253Focid%253Diehp&pid=startpage&mv=15&cv.product=prime&flightid=muidflt29-haa1t%2Cigflt3cf&exa=muidflt29-haa1t%2Cigflt3cf&activityId=1c1522b8bdb7499196bbffbdee5ea664&cvs=Browser&subcvs=homepage&evt=unload 
"Error while loading the AutoUpdater: Plugin was not installed" 
SCRIPT5007: Unable to get value of the property 'noCompleteBug': object is null or undefined 
l10n-gaia-02ca67948fe8.js, line 1462 character 5

Comment 3

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Specifically, it looks like it is this:
SCRIPT5007: Unable to get value of the property 'noCompleteBug': object is null or undefined 
l10n-gaia-02ca67948fe8.js, line 1462 character 5

Comment 4

[Mark Banner (:standard8)]

(In reply to Jared Wein [:jaws] (please needinfo? me) from comment #3)
> Specifically, it looks like it is this:
> SCRIPT5007: Unable to get value of the property 'noCompleteBug': object is
> null or undefined 
> l10n-gaia-02ca67948fe8.js, line 1462 character 5

See attachment 8504897 [details] [diff] [review] on bug 1073415. I tried working around this there. We hadn't decided how to handle patching of the l10n-gaia library, but I think we were getting towards seeing if we could get the patches into the main library (though maybe temporarily landing on our copy).

Comment 5

[Romain Testard [:RT]]

My experience using IE 10.0.9200.17148 on Windows 8:
When getting to a room URL or call URL (translation from french):
SCRIPT5007: Unable to get property  « noCompleteBug » from a null reference or non defined 
l10n-gaia-02ca67948fe8.js, Line 1462 character 5

Comment 6

[Maire Reavy [:mreavy] Plz needinfo]

[Tracking Requested - why for this release]:
This patch is for the standalone app. We may choose to uplift it for the sake of shared code with desktop.  Marking this as tracking Fx35 so we get this fixed before Fx35 goes to Release.

Comment 7

[:shell escalante]

Hi Jared,  Is this a bug you're able to do this week so we can uplift into Beta Monday for testing?

Comment 8

[Maire Reavy [:mreavy] Plz needinfo]

[Tracking Requested - why for this release]:
After triage we've moved this to blocking Fx36 (though we'd love to fix it this week).

Comment 9

[Sylvestre Ledru [:sylvestre]]

We won't block the release for this bug.

Comment 10

[Maire Reavy [:mreavy] Plz needinfo]

Moving all P1->P2.  (P2 means a major bug that we very much want to fix, but we wouldn't stop ship or block the release for it.)

Comment 11

[Maire Reavy [:mreavy] Plz needinfo]

Here's the actual change from P1->P2 (per the previous comment).  P2 indicates a major bug, but not a stop ship.

Comment 12

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Pull request for l10n-gaia, https://github.com/mozilla-b2g/gaia/pull/27606

I inverted the property name in Standard8's patch from noMutationObservers to mutationObservers, and also fixed a bug with the setter that was always using `true` regardless of the supplied value.

Comment 13

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Created attachment 8553393 [details] [diff] [review]
Patch

This patch includes the changes for l10n-gaia.

If the pull request is approved, then only the index.html changes will need to be landed.

Otherwise we will take this patch on top of l10n-gaia and have to reapply it next time we update. I introduced l10n-gaia-upstream.txt to track the changes that we are making on top of 02ca67948fe8.

Comment 14

[Mark Banner (:standard8)]

Thanks Jared,

I think you'll need to file a bug to go with the PR to get people to notice it. I did for my change - see bug 1118444.

Can you include bug 1117863 in the text file please? I've also got a patch outstanding.

There's also bug 1118905 to re-sync our version of the file once we've got these updates landed.


If we're confident these will work fine, do you think we could land this today? I'm being pushed for a roll-out of standalone for the Google analytics stuff, and so getting this released as well would be a big improvement for us.

Comment 15

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Created attachment 8553925 [details] [review]
Github pull request

Comment 16

[Zibi Braniecki [:gandalf][:zibi]]

So, we rely on mutation observers for the whole localizability on the JS->HTML API. If you want to support browsers that don't support MutationObservers, but you do want to use l10n.js, why not use a polyfill?

Disabling MO in l10n.js would require many small updates to methods like translateElement, setAttributes etc.

Comment 17

[Zibi Braniecki [:gandalf][:zibi]]

For dataset, we could instead switch to getAttribute('data-noCompleteBug'), right?

Comment 18

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Created attachment 8553971 [details] [diff] [review]
Patch v2

I got my Windows 7 VM with IE10 able to talk to my host machines server.

The attached patch gets us to show http://screencast.com/t/pPn6DUhak on IE10. I'm not sure if the MutationObservers were needed for IE9, but since this is pretty much a static page for IE I don't think we need to be concerned with having the l10n update strings as new ones get added.

I also switched to use getAttribute from gandalf's recommendation.

Comment 19

[Dan Mosedale (:dmose, :dmosedale)]

Comment on attachment 8553971 [details] [diff] [review]
Patch v2

Review of attachment 8553971 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good, r=dmose.

Comment 20

[Jared Wein [:jaws] (Regression Engineering Owner for 65) (please needinfo? me)]

Created attachment 8554596 [details] [diff] [review]
Patch v2 (with tracking file)

Carrying forward r+

Comment 21

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/fx-team/rev/46376fd5a272

Comment 22

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/46376fd5a272

Comment 23

[Paul Silaghi, QA [:pauly]]

Verified fixed on IE9 and IE10 using the dev-server.