861463 - On orientation change to portrait, document width is frequently wrong but window.innerWidth is always correct

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: General, defect)

Description

[donrhummy]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:20.0) Gecko/20100101 Firefox/20.0
Build ID: 20130326150557

Steps to reproduce:

Rotated to landscape and then back to portrait and used the following code inside a listener for window.onresize:

(onresize fires twice)
alert(
	//This alerts incorrectly as 500 and then finally 415
      Math.max( 
      	      document.body.scrollWidth,
      	      document.documentElement.scrollWidth,
      	      document.body.offsetWidth,
      	      document.documentElement.offsetWidth,
      	      document.body.clientWidth,
      	      document.documentElement.clientWidth
      ) + "," +
      
      //This alerts correctly both times as 384
      window.innerWidth
);


Actual results:

the document/body width is incorrect but the window.innerWidth is correct. I checked that the reported widths were incorrect by attempting to croll to the left or right and could not move the page, indicating that the width was exactly the same as the window's innerWidth.


Expected results:

document/body width should be correct and the same as innerWidth

Comment 1

[Kartikaya Gupta (email:[e-mail])]

When I load http://people.mozilla.com/~kgupta/bug/861463.html on my Nexus 4 running latest nightly, it alerts the expected values (980 for everything except document.body.offsetWidth, which is 964). I do get a lot of alert dialogs but that is tracked in bug 748495.

Please attach a complete standalone test case that demonstrates the problem you are seeing.

Comment 2

[donrhummy]

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #1)
> When I load http://people.mozilla.com/~kgupta/bug/861463.html on my Nexus 4
> running latest nightly, it alerts the expected values (980 for everything
> except document.body.offsetWidth, which is 964). I do get a lot of alert
> dialogs but that is tracked in bug 748495.
> 
> Please attach a complete standalone test case that demonstrates the problem
> you are seeing.

I'll put together a test to show it but I noticed a bug revealed by your code:

1. In portrait, it reports 980 and 964 as the widths (the Nexus 4 in portrait is 768 width max)

2. In landscape it reports EXACTLY the same values. This shouldn't be possible. The width in landscape on the Nexus 4 is greater than the portrait width, so why does it always report the same 980 and 964?

Comment 3

[Kartikaya Gupta (email:[e-mail])]

The CSS viewport width doesn't change, and so all the layout values (also in CSS pixels) don't change either.

Comment 4

[Kartikaya Gupta (email:[e-mail])]

Closing as WFM since the test case from comment 2 hasn't materialized and without that I can't reproduce anything that looks like a bug and matches the description from comment 0.