| Reference | Downloads | Github

Browser issues: exp runs online using Mac, but not Windows

URL of experiment:

Description of the problem:

Hi all,
After months of trial and error, and figuring out just enough JS to get by, my experiment is seemingly working and ready to go. However, after testing whether it runs properly on different computers / browsers I seem to have run into a few problems. See breakdown below:

I created the experiment on my Macbook; it runs fine there in Chrome and Firefox, all the way through to the end, including data saving. However, when running the experiment in the same browsers (and same versions) on a Windows PC, the experiment consistently freezes and/or crashes the browser. In Chrome, a black screen shows up after around a minute (regardless of which Routine you happen to be on); in Firefox, I can usually get a bit further, but after 2-3 minutes the browser will freeze and completely crash. I’ve tried to examine the JS console at the time of the crash, but Chrome is really the only one that successfully tells me anything: “CONTEXT_LOST_WEBGL”. When Firefox crashes, the console does too (even with “persist logs” turned on), though I did manage to find “EXCEPTION_BREAKPOINT” as the reason for the crash in the automatically-generated crash report.

I’m hoping there’s a simple error somewhere in my JS, though if that were the case, I can’t understand why it would run correctly in browsers on my Mac. I should also note - I don’t expect the experiment to work on all browsers, and I’m happy to explain to participants that they cannot use e.g. Internet Explorer. However, I really could do with it running across Mac and PC, as I don’t want my participants to be comprised exclusively of Mac-users.

As ever, any help would be much appreciated. I’m really eager to start collecting some of my first PhD data on Pavlovia soon! :slight_smile:

After much experimentation, I have determined this has nothing to do with particular browser or computer.

The black screen / browser crash I was experiencing was caused by having an ImageStim with an opacity set to update every frame. See: