Warning: Ignoring unknown aesthetics: x, y with geom_map

[hrbrmstr]

Before I go a-hunting and "fixing" ggalt::coord_proj() I just wanted to make sure that (a) you knew that the examples for geom_map() toss Warning: Ignoring unknown aesthetics: x, y since, well, there are no aesthetics for it anymore and, (b) that this is a desired/designed change (so I know I should compensate for it).

If this is deliberate, I'll take care of the examples for mapping in the ggplot2 package and send in a PR for you to ensure the warnings don't happen.

UPDATE

I think the mapping geom stuff may be broken.

If the x & y aesthetics are really no longer needed I should be able to do:

library(ggplot2)

world <- map_data("world")
world <- world[world$region != "Antarctica",]

ggplot() + geom_map(data=world, map=world, aes(map_id=region))

but that gives a blank map.

If I do:

ggplot() + geom_map(data=world, map=world, aes(x=long, y=lat, map_id=region))

Then I get a map and the Warning: Ignoring unknown aesthetics: x, y.

If I do:

ggplot() + geom_map(data=world, map=world, aes(map_id=region)) + coord_map()

I get:

Error in data.frame(x = x.major, y = y.range[1]) : arguments imply differing number of rows: 0, 1

(same for coord_proj() since I mirrored the coord_map() idiom very closely).

This:

ggplot() + geom_map(data=world, map=world, aes(x=long, y=lat, map_id=region)) + coord_map()

makes the usual broken mercator world map (but it renders expectedly)

This:

ggplot() + geom_map(data=world, map=world, aes(x=long, y=lat, map_id=region)) + coord_proj()

Errors out after the points are projected with:

Error in gPathFromVector(names) : a 'grob' path must contain at least one 'grob' name

which is probably on me as I suspect that there was a valid ggplot2 API change that I need to match in coord_proj() (and it may just be that some of the utility code I needed to embed in ggalt from ggplot2 needs an update).

[hadley]

It's an unanticipated change, but it is correct I believe. I honestly don't remember much about geom_map() though. I trust your judgement.

[hrbrmstr]

I added more detail from some basic tests. I have some cycles in the next 48 hours to poke at this on the ggplot2 side if you and/or @thomasp85 want an extra set of hands on this.

[thomasp85]

Much appreciated. I'm in "new job"-mode so I'll have less time in the next week or so but will be responding and advising. I'll deal with any facet related issues of course

[hrbrmstr]

Cool. I'm on this and will update prbly either very late tonight or tomorrow AM (EDT).

[hadley]

Ok, the problem with the examples is the use of expand_limits() (#1795). That's because geom_map() doesn't work in such a way that ggplot2 knows the extend of the map values, so you always have to expand it by hand yourself.

I think that means this is not a problem with geom_map() (since it hasn't changed), and is instead just a problem with expand_limits().