search

ctreffe / alfred

Alfred - A library for rapid experiment development
MIT License
10 stars 1 forks source link

Define session timeout based on last movement OR improve documentation #243

Open slussinator opened 1 year ago

slussinator commented 1 year ago

Dear Community,

I just received feedback from a participant that tried to take part in one of my studies yesterday. In the middle of the experiment they received the following message "Session expired. Sorry! the session timed out." (See attached file). The session timeout was set at 30 minutes, so this should not be the culprit. Do you have an idea what caused the problem? I remember eduroam not working well for a short period yesterday. Could this have caused the problem? Session expired.pdf

Thank you in advance!

ctreffe commented 1 year ago

Hey @slussinator,

your issue concerns an online experiment, correct? In this case, I see no reason why the short session timeout should not be to blame for the aborted session.

Either way, you can easily check whether your experiment sessions run according to the configured timeout by running a few test sessions and waiting between page changes. At this point, from a technical perspective, I see no issues or alternative explanations for the aborted session. Also, I see no way that temporary problems with an internet connection could cause the described problem.

We will, of course, revisit this issue, if you encounter any verifiable problems with the timeout configuration.

slussinator commented 1 year ago

ok. I guess maybe they just took longer than expected? Does the session timeout relate to pages or sections? Possibly it took more than half an hour to move through the section and this is why it timed out? I do understand correct though that the session timeout does not apply to the time needed to finish the whole experiment, correct? Thanks in advance!

slussinator commented 1 year ago

So I tried it out and the session timeout does apply to the time it takes to complete the study. This was not clear to me. I was under the impression that it applied to the time spent on a single page. You might want to add these specifics to the documentation.

ctreffe commented 1 year ago

This is indeed unclear from the documentation, or at least I had the same impression as @slussinator since in older versions of Alfred, the session timeout worked as she described. We should add an explanation with an emphasis on the fact that the timeout works with regard to the duration of the ongoing session and not single pages, etc.

jobrachem commented 1 year ago

I agree, clarifying the documentation in this point would definitely be valuable. Thinking further, it may also be sensible to change the behavior of the session timeout, such that a session times out with respect to the most recent activity instead of the first activity, as you expected it to. From the top of my head, I do not see any definite reason that speaks against such a change.