Heya. Can you please give a bit more detail on what the 2D browser toolbar is here? The omnibox?

yeap, it means omnibox.

Found a simple fix. So assign to myself.

correction to original report:

What happens instead?
The video keep playing. But the 2D browser toolbar shows up on top of the video. The tab switcher in the 2D browser toolbar is *NOT* tappable too.

After talk with Micheal, it looks like we are not using video overlay (the optimized path for fullscreen video) in VR and we don't persist fullscreen mode when entering VR. Supporting video overlay in VR is not trivial. So the short term solution for this issue is to exit fullscreen mode after exit VR.
This way, we are consistent with our current entering VR behavior. And we don't leave user in a non-optimized fullscreen video code path.

In the long term, we want to keep fullscreen state consistent in/out VR.
David@ none of us knows all the nuance of videos. Just want to check with you to see if this makes sense in the short term.

To the app, I guess this is the same as the user exiting fullscreen, so that should be fine. It's not ideal that the user has to go back into fullscreen, but it is probably rare that the user wanted to exit VR and continue watching in fullscreen.

What is it about video overlays that causes the observed behavior?

Would it be better to use the non-optimized path when returning to 2D (as we used to in Chrome) than to change the user's state unexpectedly? Is that not trivial? What is involved in getting the overlay to work in this case?

> about video overlays that causes the observed behavior?

i'm not sure what the omnibox has to do with overlays, either.

If one plays a video that does not use overlays (e.g., disable the feature "overlay-fullscreen-video" to turn them off entirely), does the omnibox still behave unexpectedly?

I don't really know the detail. I am assuming that it is related to the fact that we didn't use FullscreenActivity for fullscreen video in VR. FullscreenActivity does not support VR yet and it is not very trivial to support it in VR. After returning to 2D, I dont see a reason not to use FullscreenActivity for video.

To be more clear, the video overlay doesn't cause the problem. The problem looks like after we exit cinema mode from VR, the fullscreen manager still thinks that we are in fullscreen mode. But the omnibox showed up because of exiting VR. So it is in that broken state. By quiting fullscreen video after exit VR, we can fix the broken state. 

Hmm. This seem like something we should just be able to fix rather than force-exiting fullscreen. It look like FullscreenActivity was prevented in  issue 729068  because of a bug and  issue 729583  tracks fixing whatever that bug is. +peconn

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ea69a621f4528e8687a2f5d71ac5e981dcf0bc08

commit ea69a621f4528e8687a2f5d71ac5e981dcf0bc08
Author: Biao She <bshe@chromium.org>
Date: Tue Sep 26 14:59:13 2017

Don't show controls if exit VR from fullscreen mode

Bug:  767572 
Change-Id: I7a9c90372c719bc423603d9e6a9a4ce33f292191
Reviewed-on: https://chromium-review.googlesource.com/677988
Commit-Queue: Biao She <bshe@chromium.org>
Reviewed-by: Michael Thiessen <mthiesse@chromium.org>
Cr-Commit-Position: refs/heads/master@{#504371}
[modify] https://crrev.com/ea69a621f4528e8687a2f5d71ac5e981dcf0bc08/chrome/android/java/src/org/chromium/chrome/browser/vr_shell/VrShellImpl.java
[modify] https://crrev.com/ea69a621f4528e8687a2f5d71ac5e981dcf0bc08/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellTransitionTest.java

Pretty sure this is unrelated to FullscreenActivity. The real issue is video overlays not working in VR.

FullscreenActivity just doesn't really make sense for VR, and while we could support it theoretically it would probably be a fair bit of work getting VR to transition properly.

@mthiesse Do you know a bug that tracks the video overlay issue? Perhaps we can move discussion about video overlay to that bug.

Sorry for any confusion. For this particular issue, it is not related to both FullscreenActivity and video overlay. Even we fixed both FullscreenActivity and video overlay, I think we still need this fix.

The issue I am fixing here is that if we exit VR while in cinema mode (corresponding to fullscreen mode in 2D), video is still playing in fullscreen mode while the browser control shows up. We should either:
1. hide the browser control
2. exit fullscreen video mode
And we choose to implement approach 2 to fix the issue as we think it is probably more correct behavior anyway. And we also think that apparoch 2 would give user opportunity to use the video overlay/FullscreenActivity code path if they want to watch video in fullscreen while in 2D. They just need to click the fullscreen button again.

I am going to mark this issue as fixed given the reason above. Feel free to reopen if you disagree.

There's no bug tracking overlay issues, we've just disabled them when they broke us. We also don't plan on making them work for VR any time soon as far as I know.

What I meant when I said overlays were the issue is that if we were using overlays, the browser controls wouldn't be an issue as they're forcibly hidden when an overlay is present.

I just verified (in 63.0.3233.0) that when exiting Chrome VR (via the x in the upper left corner) while in cinema mode (fullscreen video in Chrome VR). After this, in 2D mode, the video is no longer in fullscreen mode, and the omnibox appears and is clickable (functioning). Is that expected? Or do we now expect the video to remain in fullscreen mode after exiting Chrome VR?

Thanks.

bshe, this sounds like it's working as intended. Could you please update this when you have a chance?

I think  issue 773417  tracks the eventual goal, but this issue is probably fixed. bshe?

Sorry for delay. That's expect behavior. mark as verified