Does this work in other browsers? YES

====================================

Good Build:

50.0.2628.0    Base Position: 370939


Bad Build:

50.0.2629.0    Base Position: 371142

=====================================

Able to repro this issue on Windows 7, MAC (10.11.4) & Ubuntu Trusty (14.04) for the Google Chrome Stable Version - 50.0.2661.75

This is a regression issue broken in M50, below mentioned is the bisect info:

CHANGELOG URL: https://chromium.googlesource.com/chromium/src/+log/e0b6a82cadecbda7e005921c2416fc7afb3542aa..1fee3b4957164505e70df4152fa31a2807fefa74

Suspecting Commit: 8876584335b48c99cf8df552ef4d8efebb131041			

Review URL: https://codereview.chromium.org/1591043002

@dgrogan: Could you please look into the issue, and if it has nothing to do with your changes and if possible please do assign it to the concerned owner.

Thank you.

I confirm this regression seems to be an unintended side-effect of the fix for issue 353580.

Unfortunately it breaks a layout technique used by many website/apps to display a fixed growable header above a scrollable area taking the rest of the screen.

Here is a (perhaps simpler) demo of this layout: http://jsfiddle.net/257zsmaz/

As mentioned on many related bug reports, this can apparently be fixed by setting "display: table-cell" on the direct child of the table-row, as shown in updated fiddle: http://jsfiddle.net/257zsmaz/1/
However this technique breaks the layout in IE9+.


@dgrogan: note that your commit 8876584335b48c99cf8df552ef4d8efebb131041 tries to mimic the behavior of Firefox, but Firefox handles both fiddles without errors: 
  - http://jsfiddle.net/pCx89/4/ (red square)
  - http://jsfiddle.net/257zsmaz/ (header+scroll layout)
Perhaps this deserves more investigation?

Thanks!

@dgrogan: Maybe I forgot to mention something obvious about http://jsfiddle.net/257zsmaz/ : the 100% height of the table-row's child is evaluated as 0 in Chrome 50, _even though_ its parent table-row (and all ancestors) have an *explicit* height set.

This should be a different situation than issue 353580, so the fix should not have impacted this case, but it appears to have done so.

The diff of 8876584335b48c99cf8df552ef4d8efebb131041 does not contain anything obvious reason, unless cbstyle.logicalHeight().isAuto() is sometimes True even when the height was explicitly set? O_o