Geom.subplot_grid with multiple layers

[Mattriks]

Hi I'm using Julia 0.4.3 and Gadfly 0.4.2. Gadfly is nice!
Here is a plot:

using Gadfly
using DataFrames

function dsinc(r, x,y) 
    z = sqrt(x.^2 + y.^2)
    z = cos(z)./z - sin(z)./(z.^r)
    return DataFrame(x=x, y=y, z=(z+1), r="r=$r")
end


randu(n, u) = u*rand(n)-(u/2) 

srand(1234)
q = linspace(-10, 10, 40)
X = vcat([[i j] for i in q, j in q]...)
D1 = vcat([dsinc(r, X[:,1],X[:,2])  for r in 1:2]...)
D2 = DataFrame(x1=randu(8,18), y1=randu(8,18),col=3.0, id=[1,1,2,2,3,3,4,4], r="r=1")
D3 = DataFrame(x2=[0.,0.], y2=[0.,0.], col2=3.0, r=["r=1","r=2"])

coord = Coord.cartesian(xmin=-10.0, xmax=10.0, ymin=-10.0, ymax=10.0)

p = plot(
    Geom.subplot_grid(coord,
       layer(D1, x=:x, y=:y, color=:z, Geom.rectbin, order=1, xgroup=:r),
        layer(D2, x=:x1, y=:y1, color=:col, group=:id, Geom.line, order =2, xgroup=:r),
# layer(D3, x=:x2, y=:y2, color=:col2, Geom.point, order=3, xgroup=:r)
),
    Guide.title(join(["r=1","r=2"], " "^48)),
    Guide.xlabel("X<sub>1</sub>"),
    Guide.ylabel("X<sub>2</sub>"),
    Guide.colorkey("z")
)

And here are some issues:

i) When using multiple layers, the facet labels aren't automatically included on the plot. I have added a workaround here by putting them in Guide.title.

ii) For the layer with Geom.line, if I remove color=, and try to instead to do:
layer(D2, ..., Geom.line, Theme(default_color=colorant"red")) then I get an error. It seems that I am forced to "apply" the color scheme in the first layer to the second layer, as I have done in the above example. Also, bizarre things happen if I try replacing col=3.0, with col="red" in DataFrame D2.

iii) It would be nice to have a Geom.segment(x=, y=, xend=, yend=) to make it easier to add multiple lines, as in ggplot.

iv) I know Gadfly suggests not to do this if you try (even if you are facetting on the same variable):

plot(
    layer(D1, ..., xgroup=:r, Geom.subplot_grid(Geom.rectbin))
    layer(D2 ...., xgroup=:r, Geom.subplot_grid(Geom.line))
)

but as long as I'm facetting on the same variable e.g. xgroup=:r, I think that including Geom.subplot_grid within the layer is a nicer syntax than the otherway around. This syntax seems more appealing (to me) in adding a layer that is not facetted (i.e.where the same geoms appear on all subplots). Here is an example.If I uncomment the:

layer(D3, x=:x2, y=:y2, color=:col2, Geom.point, order=3, xgroup=:r)

in the above code, this results in strange behaviour. But then this third layer works if I comment out the first layer (the Geom.rectbin layer)!

[Vectornaut]

You can fix issue (i) by passing the data frame and the xgroup argument to the outer plot call, as shown in the test file layered_subplots.jl, instead of passing them to each layer call separately.

However, this only works when you're using the same data frame in every subplot.

[Mattriks]

Yes for i) also see my example in #840 and scatterplot matrix in #286.