Video uploading bug

[raboody]

Describe the bug
My participants were experiencing long video-uploading times at the end of their task. Repeat experimentation suggested that this occurs about 50% of the time, and delays video uploading by 4-5 minutes. Participants reported the video uploading bar getting stuck at 0% and not moving for multiple minutes.

Experimentation suggests that the issue may have been me putting the start-recording frame before the video-configuration (exp-video-config-quality) frame. A version of the same task just changing the order of these 2 frames yielded no issues, whereas the original continued to cause issues.

I am not 100% sure this was the cause, but if it was it would probably be good to know.

Which frame or frames does it affect?
'video-config-quality', 'start-recording'

Example
https://lookit.mit.edu/exp/studies/11dd80a6-a9b9-4bca-bed0-b64f417916dd/preview-detail/

Environment (please complete):

• Lookit environment (production, staging, local + previewing or participating in study): preview & participating
• Version of frameplayer code (last_known_player_sha field on the study edit page): 51c9e97
• Operating system [e.g. Windows, Mac, Linux]: Mark's students tried many different versions
• Browser [e.g. Chrome, Firefox, Safari] and version: Mark's students tried many different versions

Additional information

See Slack conversation and Mark's students' attempts: https://docs.google.com/spreadsheets/d/1YVI7npqMWhRSpWZu8MXWzUtADgUCltn2q6vvEVVpsZA/edit#gid=1846061161

[becky-gilbert]

Update from Slack: another user has reported this problem as well, due to using a webcam-display frame during ongoing session recording.

Here's a list of frames that might cause the same problem during session recording because they display the webcam feed back to the participant):

• exp-lookit-instructions (optional parameter)
• exp-lookit-observation (optional parameter)
• exp-lookit-video-assent (optional parameter)
• exp-lookit-video-consent
• exp-lookit-webcam-display
• exp-video-config
• exp-video-config-quality

Possible solutions

At the very least we should improve the documentation to explain that certain frames are not compatible with session recording, and provide some guidance/examples for work-arounds. We do have a paragraph in the video-record mixin documentation, but it's a little hard to find, and the issue and solutions aren't laid out clearly.

Here's a list of non-mutually-exclusive options we can consider in response to this issue:

• expand the existing paragraph in the video-record mixin documentation
• add a section in the start-recording documentation
• add a new documentation page, for instance under 'Study Protocol', called something like 'Frame sequences and recording' or 'Session recording'
• add a note about this in the documentation for each of the affected frames (e.g. "This frame is not compatible with session recording. If you want to use this frame in the middle of a recorded session, you can... (add link/options/examples)").
• in EFP's parse-experiment.js, detect a problematic frame sequence and produce a console warning message to the researcher
• implement a fix in the EFP, such as detecting the presence of session recording before displaying the webcam feed and automatically pausing/stopping session recording

[becky-gilbert]

Update from Slack in response to my suggestion to stop session recording before the frame that's causing problems (webcam-display) and start a new session recording after that frame.

When I remove the webcam-display frame, the recording goes smoothly for the whole session and uploads quickly. But adding that frame into the middle of the study with the full session recording seems to derail things.
I tried adding a stop-recording frame after the first task, then showing the webcam-display, then adding another start-recording frame before the second task. When I preview the study everything seems to work for the first task, it stops recording and seems to upload the video from that task, and the webcam-display frame looks good. However, when I get to the start-recording frame for the second task I see the “establishing video connection please wait…” message for a long time (more than 10-15 minutes every time I try it). It never actually moved on to the next frame - I had to exit the study.
Instead of using start-recording and stop-recording I also tried setting the individual frames for the tasks to “doRecording”: true, but that didn’t make a difference… Initializing a second recording after the webcam-display seems to be an issue (and leads to a seemingly indefinite “please wait” message)
I’m not sure why this would be? Have other people had an issue with multiple start-recording and stop-recording frames in one session? Especially with a webcam-display frame in between?

It's not clear to me yet whether this is the same problem or a different one - we may need to open a separate issue for this.