Missing values

[earowang]

Shall we handle missing values as is now? or different?

library(dplyr)
movies <- jsonlite::read_json(
  "https://vega.github.io/vega-editor/app/data/movies.json"
  , simplifyVector = TRUE)
movies <- movies %>%
  mutate(missing = is.na(IMDB_Rating) | is.na(Rotten_Tomatoes_Rating))
movies %>%
  vega(enc(IMDB_Rating, Rotten_Tomatoes_Rating, colour = missing)) %>%
  mark_point() %>%
  config(mark = list(invalid = NULL))

[sa-lee]

The default is to filter, which I think is reasonable.
Another problem happens in the json serialisation when we create the spec from vegawdiget.
Here's the example from #27, the current serialisation doesn't include the body mass column because it's NA in R:

because it isn't explicit vega gives NaNs for the density estimate for adelie / gentoo.
If we change that to an explicit null for adelie it's area pops up

[earowang]

A new argument na.rm = TRUE gives a warning about missing values are present in makr_*().