Closed nmdefries closed 10 months ago
@dsweber2 Does this fix your plotting problem? I wasn't able to reproduce, but I was having another issue with the shiny app due to ensemble names no longer being unique. Your issue seemed similar -- 4 names being provided seems like it would be related to us generating the 4 aheads separately.
Testing shiny apps is hard in general. There are some snapshot-based tests to check that behavior is stable. We're probably not there yet, but unit tests and assertions/checks in-code would be useful. The app doesn't use so many functions, so we'd need to refactor it to isolate testable code.
Forecaster options must be unique or shiny fails in weird ways. Also check that to-be-plotted forecasters are empty even if not length-0. Sometimes we end up with a non-informative list
c("")
that doesn't trigger the empty df to be returned.