Community Cloud moved to Numpy 2. I couldn't control this because I haven't fixed package versions to try and make sharing environments easier between my local windows dev platform and streamlit community cloud which is linux. Upgrading to Np2 caused problems with pickle storage of data so required upgrading locally as well.
Also added long overdue UX tests to the the frontend of the app to provide reasonable test coverage at least of basic appearance of data and usage of each UX element.
Other changes:
Repickled MARC records using an updated version of PyMARC, fixing a bug in detecting whether something was a control field
extracted long doc strings to a seperate file
simplified import data, only serving cards with matches now with a simple ID column that prevents to create a new displayable dataframe every time a change is made. Future development may look at displaying the cards with no matches as a separate task
app now checks to see if running linux or windows to auto-detect if running on community cloud and force pulling data from S3
Community Cloud moved to Numpy 2. I couldn't control this because I haven't fixed package versions to try and make sharing environments easier between my local windows dev platform and streamlit community cloud which is linux. Upgrading to Np2 caused problems with pickle storage of data so required upgrading locally as well.
Also added long overdue UX tests to the the frontend of the app to provide reasonable test coverage at least of basic appearance of data and usage of each UX element.
Other changes: