It seems like:
I just want to plot something and don’t have time to port an entire project to
How have other people dealt with these issues?
Is there anything that works?
// i’ve tried pinning this, that and the other thing, but it seems like nothing works
PGFPlotsX works fine for me.
Is it a lot faster than PGFPlots.jl?
I thought svg’s by nature took a while to make?
edit: If it’s fast, will definitely do it. Just already had a bunch of plotting suites written for Plots.jl
Well, it depends what is slow with PGFPlots.jl (julia side or rendering side).
PGFPlots in general is not very fast so if your requirement is for the plotting backend itself to be fast, it might not be what you are looking for.
This is to be used with Interact.jl on a jupyter notebook hosted on the internet
A delay of 5 seconds when a user presses a button is probably unacceptable
Is there really no way to have Plots.jl work anymore in
edit: closest I could get (based on one of the linked posts) is
scatter(rand(10),rand(10), size=(500,300), html_output_format=:png, dpi=250)
Not sure what you are referring to here — PyPlot has all of Matplotlib’s output abilities, which means you can output vector graphics (SVG, PDF, etcetera) and any resolution that you need.
For PyPlot, when you try to increase the dpi, you end up with something like this:
Wrapping this up. It seems like
gr() is the best choice for
v0.6 at this time.
A dpi of
450 seems to give the correct scaling in jupyter (as what it used to be)
I can’t reproduce with PyPlot.jl. Using
figure(dpi=600) when plotting some random data, looks good to me, and
savefig("foo.pdf") produces vector graphics that also look good.
This should not be surprising, since Matplotlib is extremely widely used and hence is pretty solid.
I have no idea about the Plots.jl interface to PyPlot, which I don’t use, but you could try to file an issue with a minimal working example.
This seems to be a specific issue with Plots.jl and not with the individual plotting packages. I use PyPlot on both 0.6 and 0.7/1.0 and have not seen that issue so far.