Overlaying subplots when using plotProxy()/plotlyProxyInvoke() in Shiny App

I was hoping to extend the plotly.R "relayout" shiny example found here to incorporate a subplot of 3 figures with a range slider down the bottom. Here is what I tried:

library(shiny)
library(plotly)

ui <- fluidPage(
  plotlyOutput("plot")
  )

server <- function(input, output, session) {
  
  p <- txhousing %>%
    group_by(city) %>%
    plot_ly(
      x = ~ date,
      y = ~ median,
      mode = 'lines',
      type = 'scatter',
      source = 'myplot'
    ) %>%
    rangeslider(thickness = 0.05)
  
  output$plot <- renderPlotly({
    
    # Just duplicating plots for this example
    
    subplot(p, p, p, shareX = T, nrows = 3) %>%
      layout(height = 720)
    
  })
  
  observeEvent(event_data("plotly_relayout", source = "myplot"), {
    d <- event_data("plotly_relayout", source = "myplot")
    # unfortunately, the data structure emitted is different depending on 
    # whether the relayout is triggered from the rangeslider or the plot
    xmin <- if (length(d[["xaxis.range[0]"]])) d[["xaxis.range[0]"]] else d[["xaxis.range"]][1]
    xmax <- if (length(d[["xaxis.range[1]"]])) d[["xaxis.range[1]"]] else d[["xaxis.range"]][2]
    if (is.null(xmin) || is.null(xmax)) return(NULL)
    
    # compute the y-range based on the new x-range
    idx <- with(txhousing, xmin <= date & date <= xmax)
    yrng <- extendrange(txhousing$median[idx])
   
    plotlyProxy("plot", session) %>%
      plotlyProxyInvoke("relayout", list(yaxis = list(range = yrng)))

  })

}

shinyApp(ui, server)

Below is a preview of the plotly output before using the rangeslider:

And here is the resulting plotly output after I use the rangeslider:

I have also tried replacing:

plotlyProxy("plot", session) %>%
      plotlyProxyInvoke("relayout", list(yaxis = list(range = yrng)))

With this instead:

  plotlyProxy("plot", session) %>%
     plotlyProxyInvoke("relayout", list(yaxis = list(range = yrng), 
                                        yaxis2 = list(range = yrng),
                                        yaxis3 = list(range = yrng)))

...just in case I was able to independently control the individual y-axes of the subplot. However, this didn't work either. The
original example works great for the single plot; it is when I try adding in the subplots that things get whacky.

I was wondering if anyone else can confirm this plotting behaviour? Any suggestions for a workaround?

Thanks in advance! :)

[UPDATE]

I managed to find a workaround. The issue seemed to be that when calling the plotProxy()/plotProxyInvoke() functions the subplot domains got garbled :-S. However, if I explicitly set the domains for each y-axis (roughly breaking the vertical real-estate in thirds) in then everything displays as it should:

plotlyProxy("plot", session) %>%
      plotlyProxyInvoke("relayout", list(yaxis = list(range = yrng, domain = p$x$layout$yaxis$domain),
                                         yaxis2 = list(range = yrng, domain = p$x$layout$yaxis2$domain),
                                         yaxis3 = list(range = yrng, domain = p$x$layout$yaxis3$domain)))

where the object p is the original plot_ly obj that I am able to access on the server side.

Any idea why the calling plotProxy()/plotProxyInvoke() seems to garble the original y-axis domains? The workaround does solve the problem, just feels a bit clunky to have to do that each time the rangeslider updates.

Start with inst/examples/shiny/proxy_relayout/app.R, then reproduce the three-panel subplot example using subplot(), plotlyProxy(), and plotlyProxyInvoke("relayout"). Investigate why relayout changes the subplot y-axis domains, including the yaxis, yaxis2, and yaxis3 cases. Done means the rangeslider updates the y-ranges without requiring callers to restore each axis domain manually.