UserAgent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.87 Safari/537.36

Steps to reproduce the problem:
1. application contains this code: img.onclick = this.method_.bind(this, stringArg)
2. reload the application repeatedly or navigate within the application
3. click the image, look at onclick value in Chrome inspector

What is the expected behavior?
The onclick handler should be set to the bound method.

What went wrong?
Clicking the image does nothing. The onclick value shows as undefined in inspector, note this is different from the default null value for unassigned handlers.

Did this work before? Yes Chrome 49.0.2623.87 and earlier versions worked fine

Chrome version:  50.0.2661.26  Channel: stable
OS Version: x86_64
Flash Version: 

This definitely seems related to the new Chrome version. At first only one user reported the issue with nobody else on the team being able to reproduce it. Bug was filed Mar 16, local OS logs show that the upgrade to Chrome 50.x happened on Mar 10 as part of the beta channel. Now that Chrome 50 has hit stable channel, it's affecting others too.

The bug also happens in incognito windows, so it's unlikely to be related to Chrome extensions.

Adding a console.log(img.onclick) seems to prevent the issue from happening, haven't been able to reproduce it after adding that. A Chrome inspector watch that checks for assignments to the img element did show undefined values being assigned, hence the hypothesis that bind() sometimes returns undefined.

The symptoms persisted after changing the code to use img.addEventListener('click', this.method.bind(...)). The onclick property now remained null, I think this rules out the possibility that some other code was setting it to undefined elsewhere. There's no 'click' handler shown when the issue occurs, same as when I manually call img.addEventHandler('click', undefined).

Sorry, I don't have a repro case I can share publically. See http://b/27698164 for the internal bug, I can help walk through the steps needed if applicable.

I suspect this may be related to the bind implementation refactor in https://codereview.chromium.org/1542963002 , and a suspicion that the new logic may be triggering a bug in v8.

While searching for related bugs, I found  http://crbug.com/596301  "Objects constructed in another window triggering V8 miscompile(?)". Seems only vaguely similar since the symptoms and repro steps aren't the same, but that also involved bind() and things apparently started breaking starting with Chrome 50.x.

Please let me know if you need additional info from me.