So… many of our players just updated to Chrome (chromium) v53.x and now much of our core game functionality is simply not working. The game loads fine but mouse interactions and other functions simply don’t work.
I tried Google Chrome Canary which is currently using v55.x and that works fine.
Is anyone else experiencing issues with the latest stable Chrome release?
Just to note:
Using Unity 5.3.5f1 (production) and 5.3.5p5 (development). No errors or warnings in console logs.
FWIW, something seems to be causing raycasts to be invalid (i.e, .direction = (NaN, NaN, NaN)). We’re trying to isolate what is specifically causing it in code but Chrome 53 is definitely the culprit.
This sounds like a bug in Chrome’s JavaScript engine, causing it to evaluate math incorrectly. It definitely sounds like something you should file a bug report for with Google, to make them aware of the issue.
I have done this through the Chrome bug reporter. Is there somewhere else I should provide info to Google?
Im also working on a repro project to submit to Unity. Will that be helpful at all?
Do you have a link? We could ping the right people at Google.
(I report Chrome bugs here: https://bugs.chromium.org/p/chromium/issues/entry - I’m not sure if the Chrome bug reporter goes to the same place?). Be sure to specify that it is a regression and from where.
Probably not much - most likely we’d end up opening a bug report at Google ourselves for that.
Here’s an example project where the issue is occurring:
http://a.gardenacres.com.s3.amazonaws.com/webgl/ChromeBug.2/index.html
The app simply performs a Camera.main.ScreenPointToRay(Input.mousePosition) each time the button is clicked. The resulting ray data is printed in the blue box.
After a few clicks in Google Chrome v53.x you will notice that the direction data on the ray output is bogus. Works fine in Firefox (and other version of Chrome).
This is kind of a critical issue for us as it renders our game unusable for anyone using that version of Chrome.
Working fine for me on canary 55.0.2853.0 , maybe a fixe is coming?
https://www.chromium.org/developers/calendar
So we have to wait until December at the worst, assuming it’s not fixed in v54.
Sound so bad, they force us to switch on WebGL and they are not able to have something working on their web browser … Thank Google …
Oh, sorry - I meant a link to your bug report?
Thanks - I forwarded this to my contact at Google.
Thanks Jonas, I appreciate that very much!
Hello everyone, I am a member of Home3D team. we created a cloud platform that lets you create, customize and share plans. (www.planimetrie3d.com)
Now we have a similar problem to “Sirrus” or some way related to mathematics. In fact, with the latest version of Chrome V53 we encounter this problem for Windows PC only: When the character controller hit the ground, the camera flicker, and after passes through the ground.
Here I propose a demo that reproduces the problem, which occurs only on Chrome installed on Windows no with other browsers, which can be reached via the following link:
www.planimetrie3d.com/chromepcbug
We appreciate every clue helps everyone.
So, it looks like Google has bisected the problem to a specific change, and they are planning to release a patch for the issue in the next 53 dot release.
Can you verify if the problem is specific to Chrome 53, and will not happen in older or newer versions? If yes, it is likely the same. You could also submit a comment to the Chromium bug report at https://bugs.chromium.org/p/chromium/issues/detail?id=644742 to have Google take a look.
In general, it is very advisable to regularly test your content on upcoming browser versions, to be able to catch regressions like these before they ship.
It appears that this has been fixed in version 53.0.2785.113 m (just released).
Thanks.
Yes, Google reacted very fast on this one (once they had the bug report from you!), and got the patch into the next 53 dot-dot release.