allow fewer elements in named values vector in manual scales

[thomasp85]

Fix #3451

This PR relaxes the requirement on the manual scale values vector and allow it to contain fewer elements than the data. Non-existing values are implicitly considered NA and mapped to the na.value value

[paleolimbot]

This is close but I'm not sure it's exactly what you intended...I imagine that specifying values as a named vector should either match the equivalent breaks + values or limits + values, and after this PR that doesn't seem to be the case (at least with the guide). I seem to remember there was an upstream scales thing here too so maybe it's outside the control of this code. Reprexes:

CRAN ggplot2:

library(ggplot2)
#> Warning: package 'ggplot2' was built under R version 4.0.5

df <- data.frame(x = 1, y = 1:3, z = factor(c("a","b", NA), exclude = NULL))
p <- qplot(x, y, data = df, colour = z)

p + scale_color_manual(limits = "a", values = "black")
#> Warning: Removed 2 rows containing missing values (geom_point).

p + scale_colour_manual(breaks = "a", values = "black")
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.
p + scale_colour_manual(values = c("a" = "black"))
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.

^{Created on 2021-05-07 by the reprex package (v2.0.0)}

This PR:

library(ggplot2)

df <- data.frame(x = 1, y = 1:3, z = factor(c("a","b", NA), exclude = NULL))
p <- qplot(x, y, data = df, colour = z)

p + scale_color_manual(limits = "a", values = "black")
#> Warning: Removed 2 rows containing missing values (geom_point).

p + scale_colour_manual(breaks = "a", values = "black")
#> Warning: Removed 2 rows containing missing values (geom_point).

p + scale_colour_manual(values = c("a" = "black"))
#> Warning: Removed 2 rows containing missing values (geom_point).

^{Created on 2021-05-07 by the reprex package (v2.0.0)}

[thomasp85]

What is the behaviour you'd want? This change means that unmatched values gets the na value (defaults to NA which means they won't get drawn)

[paleolimbot]

It looks like the behaviour is the same but the legend is drawn differently for the named values. It looks like the behaviour and legend for limits + values and breaks + values is the same for this example, but I'd expect it to match limits + values since breaks is theoretically only for the guide (I think).

[paleolimbot]

The output is now consistent, but I think this no longer fixes the issue (all three error now).

library(ggplot2)

df <- data.frame(x = 1, y = 1:3, z = factor(c("a","b", NA), exclude = NULL))
p <- qplot(x, y, data = df, colour = z)

p + scale_color_manual(limits = "a", values = "black")
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.
p + scale_colour_manual(breaks = "a", values = "black")
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.
p + scale_colour_manual(values = c("a" = "black"))
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.

^{Created on 2021-05-07 by the reprex package (v2.0.0)}

[thomasp85]

Why am I so bad at this today...

[thomasp85]

Ah - forgot to push the fix, sorry

[clauswilke]

I've looked at the code but I have no idea why the patch does what it does. If @paleolimbot thinks it's good I'm fine with that.

[paleolimbot]

All good!

I'm not sure how deeply you want to delve into the issue...this PR is a huge improvement over the current funky manual scale, but the manual scale is still inconsistent (I would argue) with the behaviour of other discrete scales. Is it easy to have the NA be drawn (i.e., mapped to na.value) instead of removed?

library(ggplot2)

df <- data.frame(x = 1, y = 1:3, z = factor(c("a","b", NA), exclude = NULL))
p <- qplot(x, y, data = df, colour = z)

p + scale_colour_discrete(limits = "a")

p + scale_colour_discrete(breaks = "a")

p + scale_color_manual(limits = "a", values = "black")
#> Warning: Removed 2 rows containing missing values (geom_point).

p + scale_colour_manual(breaks = "a", values = "black")
#> Error: Insufficient values in manual scale. 2 needed but only 1 provided.
p + scale_colour_manual(values = c("a" = "black"))
#> Warning: Removed 2 rows containing missing values (geom_point).

^{Created on 2021-05-07 by the reprex package (v2.0.0)}

[thomasp85]

You just set na.value to something sensible, e.g. "grey"

I'm open to setting a better default that matches the default discrete scales though

[paleolimbot]

I see, you need p + scale_colour_manual(values = c("a" = "black"), na.value = "grey") to make it work. I think it'd be more consistent if na.value was set to the same thing as the default scale but no strong feelings!