Technical issues when running experiment on Pavlovia

Online experiments
1

URL of experiment: (The experiment is in running status and I still have credits. If I share the link, will this consume a credit?)

Description of the problem: I uploaded my experiment a couple of days ago to Pavlovia, after checking that it worked well on my laptop and other computers with different OS. After a couple of minutes of testing, I started receiving messages from participants (recruited via Prolific) reporting different errors. To give you some examples, two participants said they received the following error:
TypeError: Cannot read property ā€˜readPixelsā€™ of undefined

Some participants said that their browser (Chrome) crashed, and one of them even said that the experiment made his computer crash :open_mouth:. At that point I decided to pause the study on Prolific.
I am trying to understand what is wrong with the study, and see if thereā€™s a way of avoiding all these errors. The experiment is a bit long (5 mins) and has some videos. Maybe that might explain these technical issues? Also, is there a recommended ā€œlimitā€ on how many participants can take part in the study simultaneously?

2

If you share the URL of the experiment on gitlab.pavlovia.org (but not the run.pavlovia.org version) that will allow us to inspect it.

Possibilities that spring to mind are

  • a memory error where the machines are running out of memory (probably in the graphics card) while trying to load the movies. If thatā€™s it then weā€™ll need to work out how/why thatā€™s happening
  • a movie file with a different format, now supported by all browsers?

There isnā€™t a limit on the number of participants running and Iā€™m confident that wonā€™t be affecting you.

By the way, I noticed you had another post about the video not updating on each trial (sorry - I didnā€™t have any idea what might cause that and so hadnā€™t responded). Did you resolve that issue?

3

Hi @jon, thanks for your message. My gitlab experiment is here: https://gitlab.pavlovia.org/Mardock/box-movement (I changed the experiment status to Pilot mode now)

Regarding the videos, I still havenā€™t found a solution to my problem (i.e. how can I repeat a video on Pavlovia across trials, so that I donā€™t need to have multiple copies of the video for each trial?)

Edit: I just realized that other Psychopy users had similar problems a couple of months ago, here and here, and apparently the solution they found was to create multiple copies of the video filesā€¦

I think that the fact of having so many videos might be one of the main reasons why my experiment keeps crashing. Isnā€™t there a way (maybe with a code component) to repeat the video file on every trial?

By the way, my videos are all in .mp4 format.

4

Yes, repeating the video should be fine (i.e. I consider this a bug). Weā€™ll need to work out why itā€™s failing (but most likely the situation is that the lib is detecting this is the same movie and ā€œcontinuingā€ rather than ā€œrestartingā€ the video so weā€™ll need to add some explicit line to restart from beginning). Iā€™ll get @apitiot to look into both the memory and restart issues

1 Like
5

Great! Thanks a lot @jon! Hopefully this will work :slightly_smiling_face:

6

Actually, Iā€™ve just posted on your other issue, that Alain has already implemented the reset method for movies (but we werenā€™t using it in the output code) so you should be able to manually start the movies from the beginning on each trial using a Code Component, and in future weā€™ll make sure the Builder-generated code does this for you automatically

7

Great. Thanks @jon. In the meantime, and in case anyone else needs to implement the reset method, hereā€™s my previous post and jonā€™s Code Component