Narrow Bisect : 
https://chromium.googlesource.com/chromium/src/+log/41c2ecafe3e33f583e853757a2564fc1dcd5459e..5a0d86fd06f902af87107e8c5bc1564e1a1bc853?pretty=fuller&n=10000

Suspecting : r385501 from narrow bisect

@wangxianzhu : Could you please help to reassign if your change is not the cause for this issue

Deleted: Expected_iron_icon.mp4 708 KB

Expected_iron_icon.mp4 708 KB Download

Adding RB-label, please change if required.

Bisected locally and found:

commit 6a2ed97b65b691947eeffbf44ff48a8af2ddff3a
Author: khart <khart@codeaurora.org>
Date:   Wed Apr 6 10:25:08 2016 -0700

    getComputedStyle now returns used values for top, left, bottom, right

Will revert.
    
    getComputedStyle now returns used values for the top, left, bottom, and
    right properties of positioned elements, instead of returning computed
    values, if the display property is not None.
    
    R=mstensho@opera.com
    BUG= 589347 
    
    Review URL: https://codereview.chromium.org/1826423003
    
    Cr-Commit-Position: refs/heads/master@{#385493}

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f119da74988bf022c2dab4193ed289c988d51a26

commit f119da74988bf022c2dab4193ed289c988d51a26
Author: wangxianzhu <wangxianzhu@chromium.org>
Date: Wed Apr 13 19:34:11 2016

Revert of fix getComputedStyle positioned element values (patchset #4 id:60001 of https://codereview.chromium.org/1826423003/ )

Reason for revert:
Caused  crbug.com/602934 .

BUG= 602934 , 589347 

Original issue's description:
> getComputedStyle now returns used values for top, left, bottom, right
>
> getComputedStyle now returns used values for the top, left, bottom, and
> right properties of positioned elements, instead of returning computed
> values, if the display property is not None.
>
> R=mstensho@opera.com
> BUG= 589347 
>
> Committed: https://crrev.com/6a2ed97b65b691947eeffbf44ff48a8af2ddff3a
> Cr-Commit-Position: refs/heads/master@{#385493}

TBR=mstensho@opera.com,rune@opera.com,timloh@chromium.org,khart@codeaurora.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 589347 

Review URL: https://codereview.chromium.org/1888533002

Cr-Commit-Position: refs/heads/master@{#387068}

[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/animations/animation-properties-in-keyframe-are-ignored-expected.txt
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/animations/animation-properties-in-keyframe-are-ignored.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/animations/change-keyframes.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/animations/play-state.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/css/getComputedStyle/getComputedStyle-resolved-values-expected.txt
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/css/getComputedStyle/getComputedStyle-resolved-values.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/css/hover-affects-child.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/abspos-dialog-layout-expected.txt
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/abspos-dialog-layout.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/fixpos-dialog-layout-expected.txt
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/fixpos-dialog-layout.html
[modify] https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26/third_party/WebKit/Source/core/css/ComputedStyleCSSValueMapping.cpp

Why was this just reverted without any discussion? I don't think this should be happening for changes which have been in the tree this long. I think the only time we should revert patches without discussion is when they have just landed and are causing test flakiness or build issues.

The change here was being made for better compatibility with Firefox. If the breakage is just part of Chrome's UI, then it makes more sense to fix it the code there.

In any case, this isn't fixed yet. M51 has already branched so a fix will need to be merged.

Sorry, I overlooked the m51 branch email and thought we were still in m51.

We may have two choices:
1. Revert the revert, and find an owner for this bug; (I couldn't assign to khart@codeaurora.org who is not a team member yet)
2. Keep the revert and merge the revert to m51.

timloh@ which would you prefer?

Turned out that the UI script incorrectly depends on 'auto' value left and top in computed style:

    /**
     * Memoize information needed to position and size the target element.
     */
    _discoverInfo: function() {
      if (this._fitInfo) {
        return;
      }
      var target = window.getComputedStyle(this);
      var sizer = window.getComputedStyle(this.sizingTarget);
      this._fitInfo = {
        inlineStyle: {
          top: this.style.top || '',
          left: this.style.left || ''
        },
        positionedBy: {
          vertically: target.top !== 'auto' ? 'top' : (target.bottom !== 'auto' ?
            'bottom' : null),
          horizontally: target.left !== 'auto' ? 'left' : (target.right !== 'auto' ?
            'right' : null),
          css: target.position
        },
        sizedBy: {
          height: sizer.maxHeight !== 'none',
          width: sizer.maxWidth !== 'none'
        },
        margin: {
          top: parseInt(target.marginTop, 10) || 0,
          right: parseInt(target.marginRight, 10) || 0,
          bottom: parseInt(target.marginBottom, 10) || 0,
          left: parseInt(target.marginLeft, 10) || 0
        }
      };
    },

dbeam@ can you update the polymer script to conform to the new API? Thanks. Please see  bug 589347  for what are changed in computedStyle API.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f3b023bb8511c3734d1207e9cf78c61f67826af0

commit f3b023bb8511c3734d1207e9cf78c61f67826af0
Author: wangxianzhu <wangxianzhu@chromium.org>
Date: Thu Apr 14 05:57:00 2016

Reland of ix getComputedStyle positioned element values (patchset #2 id:250001 of https://codereview.chromium.org/1888533002/ )

Reason for revert:
The bug is actually caused by scripts depending on old 'auto' value.
Revert the bad revert. Sorry for the noise.

Original issue's description:
> Revert of fix getComputedStyle positioned element values (patchset #4 id:60001 of https://codereview.chromium.org/1826423003/ )
>
> Reason for revert:
> Caused  crbug.com/602934 .
>
> BUG= 602934 , 589347 
>
> Original issue's description:
> > getComputedStyle now returns used values for top, left, bottom, right
> >
> > getComputedStyle now returns used values for the top, left, bottom, and
> > right properties of positioned elements, instead of returning computed
> > values, if the display property is not None.
> >
> > R=mstensho@opera.com
> > BUG= 589347 
> >
> > Committed: https://crrev.com/6a2ed97b65b691947eeffbf44ff48a8af2ddff3a
> > Cr-Commit-Position: refs/heads/master@{#385493}
>
> TBR=mstensho@opera.com,rune@opera.com,timloh@chromium.org,khart@codeaurora.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG= 589347 
>
> Committed: https://crrev.com/f119da74988bf022c2dab4193ed289c988d51a26
> Cr-Commit-Position: refs/heads/master@{#387068}

TBR=mstensho@opera.com,rune@opera.com,timloh@chromium.org,khart@codeaurora.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG= 602934 , 589347 

Review URL: https://codereview.chromium.org/1890743002

Cr-Commit-Position: refs/heads/master@{#387246}

[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/animations/animation-properties-in-keyframe-are-ignored-expected.txt
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/animations/animation-properties-in-keyframe-are-ignored.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/animations/change-keyframes.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/animations/play-state.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/css/getComputedStyle/getComputedStyle-resolved-values-expected.txt
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/css/getComputedStyle/getComputedStyle-resolved-values.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/css/hover-affects-child.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/abspos-dialog-layout-expected.txt
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/abspos-dialog-layout.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/fixpos-dialog-layout-expected.txt
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/LayoutTests/fast/dom/HTMLDialogElement/fixpos-dialog-layout.html
[modify] https://crrev.com/f3b023bb8511c3734d1207e9cf78c61f67826af0/third_party/WebKit/Source/core/css/ComputedStyleCSSValueMapping.cpp

Rechecked above issue on dev and canary and it is still reproducible on Dev build (51.0.2704.7) but issue seems to be fixed on Latest canary build ( 52.0.2711.0 ).

here's the line in Polymer source, seems to be in iron-fit-behavior:
https://github.com/PolymerElements/iron-fit-behavior/blob/master/iron-fit-behavior.html#L150

/cc valdrin@, noms@

That line is there since the initial implementation of iron-fit-behavior :)
https://github.com/PolymerElements/iron-fit-behavior/commit/badf1004e8e77768dde3ef8c8bb9c7d5a14bcc44

I've tested the iron-dropdown and paper-menu-button in this separate jsbin, and I can reproduce the issue with paper-menu-button only (I've also disabled the animations to be sure)
http://jsbin.com/buyudob/3/edit?html,output

cc/ noms@

yfulgaonkar@  Could you please review the attached screen cast which is working fine on Win 7,Ubuntu 14.04 and Mac 10.11.4 using 52.0.2717.0.

Removing the RB label for now, please do add if its needed.

I can reproduce the issue. I've filed a bug here https://github.com/PolymerElements/iron-dropdown/issues/76

With response to comment #13:
Rechecked above issue in latest canary (build no. 52.0.2718.0) and it is working fine on Win 7,Linux (14.04 LTS)and Mac OS X(10.10.5, 10.11.4). Kindly review the attached screen cast.

Deleted: Current_behavior.mp4 863 KB

Current_behavior.mp4 863 KB Download

Since this is 'ExternalDependency'(as per C#11) hence removing the Blocker label. Feel free to add it back if someone feels otherwise.

The issue was resolved with iron-dropdown v1.4.0 & iron-fit-behavior v1.1.1
https://github.com/PolymerElements/iron-dropdown/issues/76
Please try updating these components & let me know if the problem persists.

i'll roll polymer soon to pick up the changes