I’m not at home right now, but my requirements.txt file was directly populated from the same conda environment I’ve been using to test. Look like 1.1.2 is the latest as well.
I’ll double check when I am home tonight, but not expecting it to be different as I am a very new user, so not long installed.
I just cloned your repository and ran it locally and saw the same issue as the deployed version with the multi dropdowns only allowing a single selection. Which suggests it might not be a Heroku issue. Can you check that your local version is definitely working for you with the code as it is on GitHub?
Also check that you have the exact same versions as the requirements file. If you’re on mac or linux you could do something like
So, it’s actually a problem with my callbacks which filter the other dropdown options. Before I implemented that it was working. So all the Output(ID, ‘options’).
Could anyone offer advice/examples to study on how to approach that? Seems to work but only allows one selection.
Edit: I actually wasn’t sure I should load multiple inputs and outputs together but when seeing the behaviour I was expecting I just went with it!
I had another look at the code, i think the problem is that your callbacks are filtering out all the options except for the one that is already selected. For example, if I make a selection for school the else block gets run
if not selected_school:
dfa = dfa
afa = afa
else:
dfa = dfa[dfa['School'].isin(selected_school)]
afa = afa[afa['School'].isin(selected_school)]
so now dfa contains only rows where School is the selected school. However the callback updates Output("school_filter", "options") to be [{'label': i, 'value': i} for i in sorted(dfa["School"].unique())]. Since dfa has only rows with the selected school, the only option is the school that’s already selected, so when you click on the dropdown it refuses to expand.
I suspect the other issues are all of the same nature, so you just need to double check your callback logic.