New format of calls to agg_png in ggsave break scaling function:

[andrewbaxter439]

Related to #66 - the updated format for calls of agg_png in ggsave breaks the ggsave code in @thomasp85's plot scaling walkthrough.

These commands produce the same size/resolution image:

library(ggplot2)
library(ragg)

p <- ggplot(iris, aes(Sepal.Length, Sepal.Width, col = Species)) + geom_point()

ggsave("test1.png",
       p,
       device = agg_png( 
       width = 10,
       height = 5, 
       units = "cm",
       scaling = 1,
       res = 200))
#> Saving 3.94 x 1.96 in image

ggsave("test2.png",
       p,
       device = agg_png( 
       width = 40,
       height = 20, 
       units = "cm",
       scaling = 4,
       res = 200))
#> Saving 3.94 x 1.97 in image

Whereas these work as expected (with same arguments outside of ggsave:

library(ggplot2)
library(ragg)

p <- ggplot(iris, aes(Sepal.Length, Sepal.Width, col = Species)) + geom_point()


agg_png("test1.png", 
        width = 10,
        height = 5, 
        units = "cm",
        res = 200)
print(p)
dev.off()

agg_png("test2.png",
        width = 40, 
        height = 20,
        units = "cm", 
        res = 200,
        scaling = 4)
print(p)
dev.off()

^{Created on 2021-01-14 by the reprex package (v0.3.0)}

[thomasp85]

where have you seen this "updated format" being advertised?

[andrewbaxter439]

ah good question, was just going off other issue (#66) where it's assumed to be a change. Putting the arguments to pass to agg_png outside of the brackets don't seem to work as expected:

ggsave("test.png",
       p,
       device = agg_png, 
       width = 10,
       height = 5,
       units = "cm",
       res = 100)

... produces an image 3px*1px. Putting these parts in the brackets does produce the image the correct size, but scaling argument within agg_png brackets within ggsave doesn't work as expected. Apologies if I'm misunderstanding this (and other issue is unrelated).

[thomasp85]

I'm not sure if ggsave() handles ragg correctly to be honest, but that is an issue for ggplot2, not ragg 🙂

[andrewbaxter439]

Thanks @thomasp85 - can look more into it and open a ggplot2 issue if there's still a problem. Was excited to find a way of mastering scaling plots but still trying to figure out some parts of it 😄