ggplotly export of tile geometry fails.

[paulklemm]

I don't know if I do something wrong or tiles are not yet supported by the exporter. When running

test <- data.frame(
  x = rep(c(1,3,6),2),
  y = rep(c(1,3), each = 3),
  w = rep(c(.5,2,1), 2),
  z = sample(rep(LETTERS[1:6])))

tilePlot <- ggplot(test, aes(x=x, y=y, fill = z)) + geom_tile(aes(width = w))
py$ggplotly(tilePlot);

I get the following Error:

Error in gg2list(gg) : No exportable traces
In addition: Warning message:
In layer2traces(L, df, misc) :
  Conversion not implemented for geom_tile (basic geom_tile), ignoring. Please open an issue with your example code at https://github.com/ropensci/plotly/issues

[xsaintmleux]

Thanks a lot for the example code! Your intuition was right: geom_tile is not supported yet. We'll let you know when it's done.

[paulklemm]

👍 Thanks a lot!

[drewbo]

Any idea when this will be supported or any potential workarounds? I got the same error for geom_bin2d and stat_bind2d. Thanks for the updates

[msund]

Hey @drewbo, we'd love to see your example code! One workaround could be trying out some of the native functions on Plotly? Our docs are here: plot.ly/r, and the latest for what you can do with ggplot2 support is at plot.ly/ggplot2.

[mkcor]

Hello @powernap and thanks for your patience!

Right now in Plotly your z values cannot be of type character. I'll open an issue on Plotly's end so we can support this in the future.
I can suggest the following workaround:

library(plotly)

# Your initial data frame (up to randomization...)
test <- data.frame(
  x = rep(c(1,3,6),2),
  y = rep(c(1,3), each = 3),
  w = rep(c(.5,2,1), 2),
  z = sample(rep(LETTERS[1:6])))

# Define a function to map letter back to its index in the alphabet (sorry!)
let2ind <- function(x) which(LETTERS == as.character(x))

# Create new dimension z2
test$z2 <- sapply(test$z, let2ind)

# Plot tiles for z2 instead of z
tilePlot2 <- ggplot(test) + geom_tile(aes(x=x, y=y, fill=z2, width=w))
tilePlot2  # Looks different; play with color scales?

# Send to Plotly
py <- plotly()
py$ggplotly(tilePlot2)

As of yet, we don't support as much flexibility as ggplot2 in terms of controlling per-x width:
https://plot.ly/~mkcor/475
You can edit your plot further in the GUI using the Axes button (typically, X Axis: Category vs Linear). You can change the range and spacing for the x-axis. You can change the colorscale under Traces > Style.

Hope this helps!

[paulklemm]

Hi @mkcor, thanks for this response! It looks good, I think I can work with this! Keep up the great work 👍