Closed ariekahn closed 5 years ago
Whoops, welp, yep you're right, and no it's not intentional. Thanks to these lines: https://github.com/NYUCCL/psiTurk/blob/df5a61f3ef19c296f13ccb4c6867378fc35eb410/psiturk/amt_services_wrapper.py#L293-L301
The use case is when I have, say, 3 active studies, each with their own experiment folder. When I run hit list
I only want it to show me hits related to the current folder. I wonder, what would a good alternative be?
Hit title
is available. I'll change it to also return hits that have titles that match the current title in config.txt.
Of course, the more pure fix would be to maintain a separate table with HITs for the current study, which might be easy enough also. But adds complexity.
I'd see something like the ad group as being ideal for this, but unfortunately doesn't look like that would be easily available without some cross-referencing.
Just looked at the aws docs. We could set a 'RequesterAnnotation' on each created hit. Only question would be, what to set it to. If we create a new config entry like 'study id' for the requesterannotation, and then a psiturk user creates a hit, then changes the study name, they may be surprised if 'hit list' does not show their old hits. Or maybe they'd get it easily enough. Could merge that set of hits with the set of hits that have a hitid that matches an entry in the psiturk db.
Or what else might be used to uniquely identify a study for a particular requester? Perhaps the project folder name?
And yeah ad_group
is only currently used if someone is using the psiturk ad server.
My wip https://github.com/NYUCCL/psiTurk/tree/test-suite branch fixes this by tracking new hits in a new table.
closed by 2.3.2
Spent a while trying to figure out why some new sandbox HITs I was creating weren't showing up.
It looks like HITs only show up if there's a matching participant in the database, which seems like a good idea overall, but means I have to use '--all-studies' to see HITs I just created, if no one's started on them yet. Not sure if that's intended, or what a good alternative would be?