notes.Rmd

---
title: "dev_log"
output: html_notebook
---

# Development log

## 2021-12-06

Rewriting the final-project app with a more reactive design, hopefully significantly improving performance.

Key initial fix: use a reactive variable `click_counter` to record clicks of the `Resample` button, and make the `samp_vals` array dependent on both it and the `input$samp_sd` control, so that the sample gets rebuilt whenever the control changes OR the button is clicked:

```{r, eval=FALSE}
    # initialize click_counter; the samp_vals reactive variable will
    # react to changes in it and force resample.
    click_counter <- reactiveVal(0)
    # Increment the click_counter on a click of "Resample"
    observeEvent(input$resample, {
        click_counter(click_counter() + 1) 
    })
    
    # Resample the values when the SD control changes, or when click_counter
    # changes.
    samp_vals <- reactive({
        click_counter()
        # message("samp_vals called, sd input = ", input$samp_sd)
        rnorm(SAMP_MAX, mean = 0, sd = input$samp_sd)
    })
```

## 2021-12-09

The rewrite considerably cleaned the code and has a better reactive structure, but unfortunately it didn't increase performance much. Drawing plots takes time, I guess.