layer_model_predictions won't take functions in formula argument

[nswitanek]

Having trouble specifying formula in layer_model_predictions:

Would like to do similar to:

ggplot(cocaine, aes(x = weight, y = price)) + 
    geom_point() +
    geom_smooth(colour = 'cadetblue', se = FALSE, size = 1) +
    geom_smooth(method = 'lm', formula = y ~ poly(x, 3), color = 'darkorange', se = FALSE, size = 1)

But neither of these work:

cocaine %>%
  ggvis(~weight, ~price) %>%
  layer_smooths(stroke := "cadetblue") %>%
  layer_model_predictions(stroke := "darkorange", model = "lm", formula = price ~ poly(weight, 3))

cocaine %>%
  ggvis(~weight, ~price) %>%
  layer_smooths(stroke := "cadetblue") %>%
  layer_model_predictions(stroke := "darkorange", model = "lm", formula = y ~ poly(x, 3))

[kismsu]

It's because in pred_grid there is a condition "Only know how to make grid for one variable". Hope this will change.