[FR] Series type pixel

[isentropic]

Create a new series type that draws pixels. It is similar to st=:scatter with smallest size and no border if backend cannot draw pixels natively. PyPlot and GR are capable of this where the benefit is vastly faster plotting for huge data where line plots take too long

[isentropic]

A way to start is to see https://github.com/JuliaPlots/Plots.jl/blob/master/src/recipes.jl and understand how some of the recipes work, update natively supported series and implement that series in backends that would support it natively. Else, the series should just become (fallback) a scatter plot with tiny markersize and no borders. Perhaps @LakshyaKhatri or @Diaga would be interested in this issue

[LakshyaKhatri]

Hello @isentropic, Sorry for the late response. I was focusing on CDSAPI.jl for past few days. I am starting to look at the resources provided and open a PR asap. @Diaga, I think we can work on this together.

[LakshyaKhatri]

Hello @isentropic, could you provide any link to the API documentation for pixel plots in pyplot? I tried searching for it and pyplot.imshow() is the only thing that is coming up. The other method like you stated earlier is a scatter plot with no borders and small size.

[isentropic]

Yeah with native pyplot, you could

plt.plot(rand(10), ",")

This has nothing to do with images

[isentropic]

Yes, this should be put in of the conditions, as well as supported series types in the backends.jl file. Perhaps it should only be 2d

[LakshyaKhatri]

I have implemented the recipe, but while calling the plot with st=:pixel it is falling back to the default scatter plot. I will debug it and open a PR then :)

[LakshyaKhatri]

Perhaps updating supported series will do the job 😸 . Thanks @isentropic 🙌

[LakshyaKhatri]

Hello @isentropic, can I ask what @deps does (it is there below all the recipes)? I found no resources on it under RecipeBase.jl:

Plots.jl/src/recipes.jl

Lines 124 to 146 in 5673ced

	@recipe function f(::Type{Val{:vline}}, x, y, z)
	n = length(y)
	newx = vec(Float64[yi for i = 1:3, yi in y])
	newy = repeat(Float64[1, 2, NaN], n)
	x := newx
	y := newy
	seriestype := :straightline
	()
	end
	@deps vline straightline
	@recipe function f(::Type{Val{:hspan}}, x, y, z)
	n = div(length(y), 2)
	newx = repeat([-Inf, Inf, Inf, -Inf, NaN], outer = n)
	newy = vcat([[y[2i - 1], y[2i - 1], y[2i], y[2i], NaN] for i = 1:n]...)
	linewidth --> 0
	x := newx
	y := newy
	seriestype := :shape
	()
	end
	@deps hspan shape

is it denoting the fallback series type? or something like "on which series type this recipe depends"?

If it is the prior case then in pixel recipe, it will look something like this?:

@recipe function f(::Type{Val{:pixel}}, x, y, z)
    x := x
    y := y
    seriestype := :pixel
    markersize := 1
    markerstrokewidth --> 0
    ()
end
@deps pixel scatter

[isentropic]

You need to add an entry in recipes.jl I believe