FindingFive Discussion Forum
Studies featuring many audio and video stimuli may not run in Chrome 92 - Printable Version

+- FindingFive Discussion Forum (http://discuss.findingfive.com)
+-- Forum: Announcements (http://discuss.findingfive.com/forumdisplay.php?fid=7)
+--- Forum: Known Issues (http://discuss.findingfive.com/forumdisplay.php?fid=10)
+--- Thread: Studies featuring many audio and video stimuli may not run in Chrome 92 (/showthread.php?tid=129)



Studies featuring many audio and video stimuli may not run in Chrome 92 - Ting - 07-27-2021

Dear Researchers,

Thanks to a timely report from the LEAP lab at University of Arizona, we have discovered a new bug in the latest version of Google Chrome (v.92) that has impacted some of FindingFive's functionalities. Studies featuring many audio and video stimuli, including audio recordings that result from the audio responses, may become stuck at certain point, with an error message from FindingFive warning about missing stimulus file or incorrect format. Since this is a browser-level bug, your currently active sessions may be affected as well.

The cause
This is a new bug introduced by Google. Google is already working on a fix (https://bugs.chromium.org/p/chromium/issues/detail?id=1144736#c27).

Actions on FindingFive's end
We are also working to identify a workaround. However, since this is a browser-level bug, a fix might not be trivial. We will update everyone as soon as we have more insights into this issue.

What you can do
If your study features many audio and video stimuli, or use audio recording responses, please preview them immediately to see if you are impacted. If so, you can ask participants to use Firefox instead of Google Chrome to complete your study until either Google or FindingFive implements a fix.

Thanks for your patience and understanding!


RE: Studies featuring many audio and video stimuli may not run in Chrome 92 - Ting - 07-27-2021

We have identified solutions to work around Google's annoying new limitations. However, due to the complexity of this issue, we estimate that it will take us at least two weeks to implement this workaround. Hopefully Google will fix this bug before then.

At this time, the only workaround is to ask your participants to use Firefox if your study is affected. We are sorry about this inconvenience.


RE: Studies featuring many audio and video stimuli may not run in Chrome 92 - Ting - 07-28-2021

One more update - this bug has caused quite a bit of upstir among web developers, and Google has heard our feedback: https://bugs.chromium.org/p/chromium/issues/detail?id=1232649

It now appears that Google will temporarily put a band-aid on this issue in an upcoming update to Chrome on Aug 5. We'll monitor this issue and keep everyone updated.


RE: Studies featuring many audio and video stimuli may not run in Chrome 92 - Ting - 08-03-2021

The FindingFive team is delighted to announce that effective immediately, studies with audio stimuli, video stimuli, clickable videos in choice responses, and audio responses will function normally even under the problematic Google Chrome v92. We came up with our own workaround that "circumvents" Chrome's current bug (read on for details).

What is this about?
On July 20, Google released Chrome v92 (which remains the most up-to-date version), which unfortunately contains an odd, poorly-communicated, and arbitrary limitation on the max number of audio or video elements on a single web page (as an attempt to address a memory-leak issue). As the limit is set at 75, a fairly low number, it broke many web applications, including some FindingFive's studies.

What will Google do?
Google has regretted this sudden change and agreed to reverse course. However, Google's update/fix will not be available till at least the end of this week. For Chromebooks, the fix may not come until mid-August.

What did FindingFive do?
FindingFive recognized the urgency of this issue due to the huge market share of Google Chrome. We decided that it is not acceptable to wait for Google to fix this problem, because there will always be some participants who may not be able to upgrade from Chrome v92 in the near future. Therefore, we started immediately working on a mitigation solution on July 28.

We are happy to report that we have been successful at implementing a workaround for Chrome v92's odd behavior. This workaround applies automatically to all studies.

What do researchers need to do?
As the workaround results in a fairly major upgrade to FindingFive' Study Grammar system, researchers will need to preview their studies again before being able to launch new sessions that work under Chrome v92. Importantly, please do not select "Launch a new session using a previous version of the Study Grammar", as none of the previous versions will work under Chrome v92.

Buggy? Please let us know!
We burned quite a bit of midnight oil to implement this workaround, so it might be a tad buggy. Existing sessions won't be affected by the workaround, so there's no worry there. But if you notice suboptimal behavior when previewing your study, please feel free to chime in on this post!