Hi Daniel... ambitious but I think it will be incredibly useful (even if 
I'm the only one that ever uses it).  Did you see my last post about 
PyPlot?  I was working on PyPlot support yesterday, and you can do a ton of 
stuff already.  See the 
examples: 
https://github.com/tbreloff/Plots.jl/blob/master/docs/pyplot_examples.md

Matplotlib's subplot command is very clunky and I haven't figured out the 
right way to implement yet, but it already has support for:

   - Many lines, styles, symbols
   - Bars, histograms, heatmaps
   - 2-Axis support  (which was always so annoying to do... I can't tell 
   you how many times I looked at this example 
   <http://matplotlib.org/examples/api/two_scales.html> throughout my 
   career...)


In reference to:

>  The priority for me is to be able to fiddle with the details of the plot: 
> change the font, define a new colour, remove the tick marks, have two 
> y-axes, change the aspect ratio, insert formulas in LaTeX, etc.


You can already pass in arbitrary colors, and change the sizes, etc.  If 
there's functionality you can't accomplish through Plots, you can always 
adjust it directly:

using Plots
pyplot!()

# use Plots to create a base figure
plt = scatter(randn(1000),randn(1000), background_color = :red)

# say you want the background color outside the canvas to be a different 
color than inside (can't do this directly right now)
fig = plt.o.o   # fig is the PyCall wrapper around the matplotlib.Figure 
instance
fig[:set_facecolor]("blue")

# refresh
display(plt)

which displays a scatter plot with a blue exterior background and red 
interior background.

Long story short, you get all the benefits of Plots, and you can still 
access all the functionality of matplotlib.  Please try it out and open 
issues if you find yourself wanting features or if you find bugs.  And once 
you're comfortable, try out some other backends!  You might find Gadfly 
plots look great when you have trouble with matplotlib, or vice versa. 
 Also I recommend trying the UnicodePlots backend... it's not as 
feature-full as others, but it has minimal dependencies and is pretty cool!

Reply via email to