[Elise]

Hi Jeremy,

Thanks for looking into this!

I tried using the new version, but I think it happened again. I got this error:

server.cgi?include=js:formatted:3351 Uncaught (in promise) TypeError: Failed to fetch

This is where the error occured in the code: https://pasteboard.co/6ampth46pnN9.png

Since I know that the rest of my experiment is stable under the 2.0 version, I went back to it. I’ll downsample my audios and just hope that the error doesn’t occur that often.

Best,
Elise

• This reply was modified 2 years, 1 month ago by Elise.

[Elise]

Hi Jeremy,

Sure! It’s this one:

https://oltrogge.experiments-goethe.de/exps/Smoothie/main/experiment.html

Best,
Elise

[Elise]

Hi Jeremy,

Thanks for the reply! I changed to preloading the ressources from zip-files.

I am running into another issue where sometimes the experiment crashes. I get the following error in the console:

Uncaught (in promise) TypeError: Cannot set properties of null (setting 'controls')
    at c.window.PennController._AddElementType.uponCreation (server.cgi?include=js:formatted:11140:43)
    at c.n.uponCreation (server.cgi?include=js:formatted:3254:43)
    at server.cgi?include=js:formatted:3294:64
    at new Promise (<anonymous>)
    at Array.<anonymous> (server.cgi?include=js:formatted:3294:34)
    at server.cgi?include=js:formatted:1034:39

I can see that this relates to an audio ressource. Evidently, the ressource .object is null at the time. When this happens, the experiment crashes without any possiblity of recovering. This only happens once in a while and so far I have not find a way to reproduce it consistently. I’ve seen this thread (https://www.pcibex.net/forums/topic/typeerror-error/), but I checked that all my files are uploaded and named correctly.

So again, I wonder whether the new version has fixed this issue, because I see that the code relating to preloading has changed. But since I cannot reproduce the error consistently, there is no way for me of knowing whether the problem has already been solved. If you could give me some confidence, that would be great!

Best,
Elise

• This reply was modified 2 years, 1 month ago by Elise.

[Elise]

Hi Jeremy,

Sorry, it was indeed my typo. Thank you, it works now on the external server.

Best,
Elise

[Elise]

Hi Jeremy,

I noticed that without DebuggOff() I cannot switch off the debugger when running my experiment on an external server. I assume I would need to use an old version of PCIbex to make DebuggOff() usable again. Could you point me to the .js file I would need to replace by an old one such that I can use DebuggOff() again? Or can you think of any other way to switch off the debugger when running experiments on an external server?

Thanks in advance!

Elise

[Elise]

Hi Jeremy,

I see, thanks a lot! This makes total sense and the shortcut will be really helpful to me!

Best,
Elise

[Elise]

Hi Jeremy,

Regarding the callback on the mousetracker I found the reason why it wasn’t working for me: I accidentely had activated the touch emulation. Make sense that there wouldn’t be any mouse movements with touch. With the callback working I think I can now implement the behaviour that I aim for.

Best,
Elise

[Elise]

Sorry, something went wrong with the link. Here it is: Experiment

[Elise]

Hi Jeremy,

This is great solution! Thanks a lot!

Happy holidays,
Elise

[Elise]

Just saw your reply. Thank you so much! I’ll only create a single global_main.css, I didn’t know that the global_z.css is not necessary anymore.

Best,
Elise

[Elise]

Hi Jeremy,

I think I catched the error: It seems like sometimes what is defined in the global_main.css (e.g. the body tag) is executed before what is defined in the global_z.css (see https://pasteboard.co/4nkpZfZNQr1Z.png) and sometimes it is executed the other way around (see https://pasteboard.co/TCMgkkbP2CCa.png). I could solve the problem by making the tag more specific.

Best,
Elise

[Elise]

Hi Jeremy,

Thanks for the fast reply! I added the semicolon, but I can still reproduce the bug after reloading a few times. I commented out the first instruction screens such that the different style is better visible: It is mainly about the position of the stimulus / html screens. Sometimes they are positioned higher than other times.

Best,
Elise

[Elise]

Thank you very much, Jeremy! It worked perfectly!

[Elise]

Hi Jeremy,

Thanks for the fast reply!

I couldn’t use .log(“all”), because when no key was pressed the participant had to confirm a timeout message with another keypress. So instead of getting a “Never” in the results file, it saved the keypress from the timeout message.
Anyway, I now managed to implement the workaround and it works perfectly fine.

Thanks a lot for your help!

Elise

[Elise]

Hi Jeremy,

I have the same problem as Kayla. I also cannot use .log(“first”) or .log(“all”) for the same reasons and I additionally need the script to report when no key is pressed.

Is there already any fix? I am not sure how to implement the alternative that Kayla proposed. I would really appreciate any help!

Elise

[Author]

Posts