What do you mean by edges are visible? I don't see anything broken when I try on a recent chromium build. (I haven't checked M58)

vr.with.in doesn't use the controller, so that's expected not to work, and the content is supposed to follow head movement (or am I not understanding you correctly?)

I'm also not seeing anything wrong on Chrome Dev (58.0.3029.11).

Keep in mind that vr.with.in is *immensely* slow, so we will miss lots of frames, and the viewports will not stay centered as you move the screen around (is this what you mean by the edges are visible?)

By edges are visible I mean that when I move my head slowly, I can see the black edges of the world (for lack of better terms) for a second or two on the side of the screen where my head is turning towards. Before, it would not do this. Especially with slow head movement. So if this is what you mean by the site being *immensely* slow, it definitely wasn't slow last week when I tested it. It always seemed pretty snappy to me.

This site has used the controller until now. The site shows a grid of different videos that you can select with the controller action button and play pause and exit them and navigate the menu. I've used it many times before. 

The content that moves around with head movement now is the grid of videos that has been fixed in position before (last week when I tested it). Not it moves around a little bit.

In addition, the origin trial used to be enabled on this site, so we weren't required to flip the webvr flag. But now we do have to.

Basically the site was working very well, quick, and I could select and play content with the controller before. Now the site is unusable, controller doesn't work, etc.

 issue 698923  fixes rendering performance.  issue 701444  fixes the controller not working.

I think once my fix for 698923 lands we can consider this fixed.

This has somehow resurfaced with 58.0.3029.76

Is it a site issue?  Can we verify that it works on the previously tested version?

Looks like a site issue. Just tested with last week's version which worked before, 58.0.3029.66, but now it doesn't. 

The issue is a bit different now with:
Chrome version 59.0.3071.24, 
VR Core Production version: 1.4.151195927

Now it looks good until you hit the controller button. Then the screen freezes, head movement works, but the screen doesn't update. So if you look around, you end up looking at black content.

Pinged within to take a look.

Taking a peek now. We made a number of changes over the last couple of weeks for other platforms, so it is certainly a possibility of something on our end. We now have the latest origin tokens in place.


My Chrome Stable (57) seems slow and low-res on my pixel, but operates (selecting films, watching films, changing resolution, etc.). Chrome Beta (58) has rendering and performance improvements. I can verify that Chrome Canary (59) crashes for me. I believe this is something on our end. I'll keep you guys posted as I continue to diagnose.

Sorry, version updates here: Chrome Stable (57), Chrome Beta (59) and Chrome Canary (57, maybe that's why Canary is currently crashing). Updating all my versions of Chrome and will report back.

Okay, I've made updates to https://vr.with.in/ and should be working in M-59 now. When did touchpaddown and touchpadup start spoofing mouse events?

FYI: M-60 Chrome Canary still crashes on load.

Thanks, jono.

dbbrooks: Please check again.
If you see a crash as in #14, please file a bug for that.  Issue 718501  tracks a crash on a with.in page, but it's not clear if it is the same one.

works with 59.0.3071.39