support scalars for hline! and vline!

[fkastner]

It would be nice to have hline!(0.73) working instead of hline!([0.73]). The same goes for vline!.
Currently for floats it errors with No user recipe defined for Float64.
For Ints on the other hand I get a BoundsError attempt to access 0-element Array{Float64,1} at index [1].

I am 70 percent sure this worked about four weeks ago but if not then consider this a feature request :)

[yha]

Maybe we should just generally have @recipe f(x::Real) = [x]?
That way you can also do e.g. scatter!(3,5) to add a single marker to a plot.
This idea conflicts with the current interpretation of integers as adding empty series (plot(n) adds n empty series). But is this really useful?

I am 70 percent sure this worked about four weeks ago but if not then consider this a feature request :)

I remember this being an annoyance for quite some time. Perhaps somehow it was temporarily working recently, but I doubt it.

[daschw]

This idea conflicts with the current interpretation of integers as adding empty series (plot(n) adds n empty series). But is this really useful?

It's used in the Lorenz Attractor example on the start page of Plots Docs, http://docs.juliaplots.org/latest/#simple-is-beautiful, to initialize an empty series and grow it in an animation. I don't know how much it's used by users, I don't use it. But I think this change requires some discussion as it would be breaking.

I remember this being an annoyance for quite some time. Perhaps somehow it was temporarily working recently, but I doubt it.

I also cannot imagine that this worked recently. but I am not 100% sure.

[AshtonSBradley]

Does this really break anything? Can't we just set a generic scalar fallback for hline!, vline! of the form

hline!(x) = hline!([x])

so the previous behavior is retained.

In

http://docs.juliaplots.org/latest/#simple-is-beautiful

they use vline!([a])

so no breakage. For the Int problem above, why not just do a quick convert to Float? its all Float32 on the plot anyway, right? I don't see why this has to involve scatter, that presumably has its own use cases.

Isn't this just a one liner to make a rational fallback on top of existing behaviour?

I dont see why this is relevant

Maybe we should just generally have @recipe f(x::Real) = [x]? That way you can also do e.g. scatter!(3,5) to add a single marker to a plot. This idea conflicts with the current interpretation of integers as adding empty series (plot(n) adds n empty series). But is this really useful?

I am 70 percent sure this worked about four weeks ago but if not then consider this a feature request :)

I remember this being an annoyance for quite some time. Perhaps somehow it was temporarily working recently, but I doubt it.

This seems to be about new plots.