Plots slows down entire Jupyter notebook

[Bebotron]

I have 4 files that I load with readdlm, each about 5~6KB each. When I plot them using Plots (they plot successfully), it slows down the entire notebook. I can basically edit cells for 3 lines (exactly) just fine, and then any more lines are incredibly laggy. Has anyone else experienced this as well? I thought the problem might be readdlm, since that has some issues but only with very large data. My files are not that big and running a cell only loading the data works just fine. Additionally, plotting very simple things also don't cause much of a problem:

x = collect(0:0.01:2*pi)
y = sin.(x)
plot(x,y)

This doesn't cause any problems. So is it a problem of Plots struggling with larger data?

[daschw]

Does it help if you add default(fmt = :png) after using Plots?

[Bebotron]

How interesting, it does fix the problem. However, I am using a dark theme on jupyter, and so I've been using theme(:juno). So if I do this:

using Plots
using PlotThemes
theme(:juno)
default(fmt = :png)

And this will fix any lag, but will now make the plot axes and labels dark, which makes them hard to read with the dark background. So if I switch the order around to:

using Plots
using PlotThemes
default(fmt = :png)
theme(:juno)

And now the axes coloring is correct, but the lag is still there. Might there be something in the juno theme that is setting the default formatting?

[daschw]

theme(:juno, fmt = :png)?

[Bebotron]

No that still makes the axes and labeling dark. I'm trying to find the theme files and see if there's something obvious there.

[daschw]

Hm the issue here seems to be that setting a fontcolor does not work for GR:

plot(rand(10), xguide = "Hello", tickfontcolor = :blue, guidefontcolor = :blue)

[Bebotron]

Yes, setting fmt = :png within the juno.jl specifications gives dark axes.

[daschw]

When I plot from the REPL setting the font color works:

plot(rand(10), xguide = "Hello", tickfontcolor = :blue, guidefontcolor = :blue)

When I save this plot as PNG I get

That's why setting fmt = :png messes up the theme in GR.
cc: @jheinen any ideas?

[Bebotron]

Using the pyplot backend works just fine. So in case someone just needs a work-around, just do

using Plots
using PlotThemes
theme(:juno)
pyplot()

It would be nice if this issue could be fixed for Plots, but for now this is something anyone else who might be experiencing this could use.

[daschw]

Saving as SVG should also work, but leads to the original lagging issue in jupyter for plots with a lot of data.

[jheinen]

The color problem can (hopefully) be fixed - the SVG problem probably not. What does "a lot of data" mean (MWE)? Plotting millions of points makes only sense when using GR's shading routines.

[Bebotron]

In my particular case, I only have 200,000 data points to plot, which definitely isn't much. I don't know at which specific point the plotting starts lagging the entire notebook, but my particular problem can be emulated with the following:

using Plots
x = collect(0:1e-4:20)
y = sin.(x)
plot(x, y)

After generating this plot, the entire notebook is laggy.

[jheinen]

In this case, you should set gr(fmt=:png). Most browsers can't handle such point arrays in SVG.

[Bebotron]

Yes, and like we discuss above this fixes the lagging issue, but then the axes coloring doesn't work (remains black).

[daschw]

The color problem can (hopefully) be fixed

The color problem is in GR, not in Plots, right @jheinen ?

[jheinen]

I get the same results for pyplot() and gr(fmt=:png). Am I missing something?