Closed lindahua closed 9 years ago
Glad you're liking
I actually started doing this and then stopped since I wasn't sure if the ggplot2 interface is appropriate: https://github.com/johnmyleswhite/VGPlot.jl
I'm right in the middle of rewriting the types used in Vega.j to avoid code duplication, which will make it easier to track changes to the raw Vega library's types. That work should be done midweek, after which I'll try tackling the grammar-style interface once again.
I'm going to close this. I've reached reasonable feature parity with the Python Vincent library, and the Julia code has become very Pythonic along the way. As of now, each visualization has very few keyword arguments, then I'm creating mutating functions for the behaviors like so:
using Vega
x = [0,1,2,3,4,5,6,7,8,9,0,1,2,3,4,5,6,7,8,9]
y = [28, 43, 81, 19, 52, 24, 87, 17, 68, 49, 55, 91, 53, 87, 48, 49, 66, 27, 16, 15]
g = [0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1]
b = barplot(x = x, y = y, group = g, stacked = true)
colorscheme!(b, ("Greens", 3))
As it is, I'm not really a fan of the gg syntax, as theoretically sound as it might be.
Good call. Shrinking the scope of this package is a good way to go.
I am really excited about Vega. This is a framework that is user friendly and expressive.
What remains needed is a way to express the grammar that really unlocks the power of this grammatical system (just like ggplot2).
Considering the plot in http://trifacta.github.io/vega/editor/ (the scatter plot)
it would be nice to be able to express that as
Here,
vgplot
is similar toggplot
in ggplot2 of R. Also, thesexaxis
,yaxis
,points
andlegend
should be optional. If they are not specified, the function should have a reasonable default behavior.