saving and restoring state

Are there plans to add tooling to save and restore state in py-shiny? I'm particularly interested in the option to save state to files (pickle format?) and to databases (e.g., sqlite). For apps that use large-ish datasets allowing the combination of storing to file + database would be ideal.

@jcheng5 got me started in an approach (re)store state in shiny-for-R with https://github.com/jcheng5/shiny-resume/blob/master/session.R. This required functionality like parseQueryString(session$clientData$url_search) which I don't see in py-shiny as yet.

The approach that I have used relies on functions that control how state is stored and restored for each input. See https://github.com/vnijs/quizr/blob/master/inst/example/init.R for an example. r_state is a list that is populated on session$onSessionEnded as shown below.

saveStateOnRefresh <- function(session = session) {
  session$onSessionEnded(function() {
    isolate({
      r_sessions[[r_ssuid]] <- list(
        r_state   = reactiveValuesToList(input),
        timestamp = Sys.time()
      )
      ## saving session information to file but could also be store to a database
      ## storing state could also be initialized using a button
      saveRDS(r_sessions[[r_ssuid]], file = paste0("~/r_sessions/r_", r_ssuid, ".rds"))
    })
  })
}

If a value is available in the r_state state list it will be used to initialize inputs on restore.

state_init <- function(inputvar, init = "") {
  if (is.null(r_state[[inputvar]])) init else r_state[[inputvar]]
}
radioButtons("radio1", NULL, answers, selected = state_init("radio1"))

I'd be very interested to hear your thoughts on (re)storing app state in py-shiny

Start by comparing the shiny-resume session.R example and the quizr init.R example, then review the py-shiny session lifecycle around session$onSessionEnded and the missing parseQueryString equivalent. Define the persistence scope, supported file or database backends, and restore behavior before implementation; done should include a documented state-saving and state-restoring workflow.