`histogram(density = true)` and `x_discrete(labels = ...)` incompatible ?

[EliasBcd]

I am trying to plot a histogram while giving labels to each histogram bar. There seem to be an incompatibility between density = true and x_discrete(label = ...)

x = [1,1,1,1, 2,2,2,2,2]
lab = Dict(1 =>"one", 2 => "two")
Gadfly.plot(x = x, Scale.x_discrete(labels = x -> lab[x]), Geom.histogram(density = true))

On my computer, the density = true is ignored when I have x_discrete, but works perfectly without latter. It looks like a bug to me, or is it a normal behavior? Thanks!

[bjarthur]

@tlnagy does #1157 change this behavior?

@AsileBcd this works for me:

plot(x = x,
    Scale.x_continuous(labels = x -> lab[x]),
    Geom.histogram(density = true),
    Guide.xticks(ticks=collect(keys(lab))))

[tlnagy]

#1157 won't touch this behavior. I don't fully understand this use case. @AsileBcd could you describe why you're trying to do this and not using Geom.bar?

[Mattriks]

Another example that illustrates the issue. I think there is a case that discrete variables can be proportioned, what do people think?

x_cont = [1,1,1,1,2,2,2,2,2]
x_disc = string.([1,1,1,1,2,2,2,2,2])
pa = plot(x=x_cont, Geom.histogram(density=true))
pb = plot(x=x_disc, Geom.histogram(density=true))
hstack(pa, pb)

[EliasBcd]

Sorry for the late answer.

@tlnagy : My little example is a little bit silly obviously. The general idea is inline with @Mattriks idea: instead of computing the proportion before with a count function, being able to represent directly the proportions (and potentially switch easily between counting and proportional representation). I also just think that the behavior is not the intended and should be documented as such.

@bjarthur It also works for me, but if I am picky, the alignment of the labels is not right.

x = [1,1,1,1, 2,2,2,2,2]
lab = Dict(1 =>"one", 2 => "two")
pa = plot(x = x,
    Scale.x_continuous(labels = x -> lab[x]),
    Geom.histogram(density = true),
    Guide.xticks(ticks = collect(keys(lab))))
pb = Gadfly.plot(x = x, 
    Scale.x_discrete(labels = x -> lab[x]), 
    Geom.histogram(density = true))
hstack(pa, pb)

[bjarthur]

you can hack around it with

pa = plot(x = x,
           Scale.x_continuous(labels = x -> lab[x-0.5]),
           Geom.histogram(density = true),
           Guide.xticks(ticks = 0.5+collect(keys(lab))))