psychopy.org | Reference | Downloads | Github

Pavlovia Committer deleted my entire repo when starting data collection

URL of experiment: gitlab.pavlovia.org/floelling/ambiguous-future.git

Description of the problem:
I had a full experiment running and sent the URL link out through SONA. For some reason the Pavlovia Committer then deleted all of my files. and replaced it with a dummy participant data file. Erasing the resources I was using for the experiment. Somehow the study was still able to run a couple of subjects but now I get errors that the resources I download mid experiment are missing. I know I can revert the project back to the previous commit and restore it to its previous state. But I’m not sure how to handle this if it happens again and how to make sure to keep my data. Thanks for your help!

This should happened again to me! I would really appreciate any feedback you have on this. I’m at a loss for what to do here.

Thanks for spending your time on this!

Hello @kdoelling,

As per the email you separately sent me, this might have to do with your issuing git commands independely from PsychoPy and the conflict that might have resulted from it. Did you once delete those files and committed that change?
Would it be possible for you to generate a new experiment and check whether you observe the same behaviour?
As per your email, I would very much appreciate if you could indeed limit the size of your wav samples: 4.1GB is really rather a lot. Do they have to be in such high definition?

Alain

Thanks for your response Alain! I appreciate you being generous with your time. I am working now on a new repository which syncs only through psychopy. However it has inherited the commit history from before. Do you think that could be a problem? I haven’t ever deleted the entire repo and committed the change. At least going through the logs I don’t see any evidence for something like that. Here is the new repo: https://gitlab.pavlovia.org/floelling/future-uncertainty.

My repo (when things don’t get multiplied through reverts and such) is showing itself to be 300 MB. Do you think 300 MB is also too large? The repo in question with the commit bloat says that it is 1.2 GB. Once the new repository works, I will erase this repository to make space. If there is another location where I have inadvertently taken up 4.1 GB of storage, please let me know and I will get rid of it immediately.

The reason we have so many wav files is that our experiment hinges on having different files for each subject. In python or matlab, we would generate each trial on the fly based on a pattern of tones drawn from a noisy distribution. We couldn’t figure out how to do this in javascript so we instead generated a large number of sound files in advance and draw randomly from them for each participant. If you think there is a better solution I would be happy to change it in the service of storage reduction!

Thanks again for your help!

I am a little surprised about your mentioning that the new project has inherited the commit history from the previous one. Did you fork it?

In terms of space: future-uncertainty/html/wavs occupies 4.1GB…

OK, I can see that you are 100% right. For some reason on the front page of the repo it says “301.3 MB Files” so i assumed that was the entirety of the repo. But maybe it doesn’t count wav files which it cannot easily read.

With regards to the commit history, I used the same builder file to create the new repository and sync with it. I think because the folder the file was in had the same .git folder in it, the push command took everything with it.

Would you suggest that I:

  1. delete the current future-uncertainty
  2. downsample my wav files
  3. delete my local .git folder
  4. create a new future-uncertainty repo and sync with it through psychopy3

Do you think creating a new version of the repo with the same name could create issues? I.E. should I come up with a new name for the repository?

Thanks again for your help

For the record: This worked and it was effective. I am marking it as the solution. I’m fairly certain all of my issues were the result of a repository that was just way too large. Thanks for your help @apitiot.

I am very glad that it worked out in the end.
Happy experimenting!

Alain