The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/18f047382923518df79038a3a58cc7f8ecc0c712

commit 18f047382923518df79038a3a58cc7f8ecc0c712
Author: dpapad <dpapad@chromium.org>
Date: Mon Mar 21 22:48:52 2016

MD Settings: Make <settings-dialog> respond to ESC key.

BUG= 596637 

Review URL: https://codereview.chromium.org/1822813002

Cr-Commit-Position: refs/heads/master@{#382423}

[modify] https://crrev.com/18f047382923518df79038a3a58cc7f8ecc0c712/chrome/browser/resources/settings/settings_dialog.js
[modify] https://crrev.com/18f047382923518df79038a3a58cc7f8ecc0c712/chrome/test/data/webui/settings/cr_settings_browsertest.js
[modify] https://crrev.com/18f047382923518df79038a3a58cc7f8ecc0c712/chrome/test/data/webui/settings/reset_page_test.js

I am re-opening this bug since it has regressed again. This time the cause is different. Bisection showed range [1], and I narrowed it down manually to change at [2]. 


[1] https://chromium.googlesource.com/chromium/src/+log/e731d94e5654cb809ecc093f08736f8984a6ffd6..118b85eb19b9761a27acd727fa03dfce5d00bba3
[2] https://chromium.googlesource.com/chromium/src/+/a02ef8b2b472ef649f8881dfb52f97676c9bc3b4

The fact that the test added before is not failing, makes me think that this time something changed regarding the keyboard event that is being fired by Chrome (since the test is ensuring that given an ESC event the dialog is closing). Needs further investigation.

An easy way to reproduce this locally, is to visit https://elements.polymer-project.org/elements/paper-dialog?view=demo:demo/index.html&active=paper-dialog. Open any dialog and hit the "ESC" key.

This is a problem in Polymer and Chrome will re-run crisper after their fix.