Adding 7 new markers

[jonathanrocher]

The new markers proposed are: left and right triangle, star, cross-plus, hexagon1 and hexagon2, and pentagon.

To try them out, you can run:

import numpy as np
from traits.api import HasTraits, Instance
from traitsui.api import View, Item
from chaco.api import Plot, ArrayPlotData
from enable.api import ComponentEditor

class MyPlot(HasTraits):
    plot = Instance(Plot)
    view = View(Item('plot', editor=ComponentEditor(), show_label=False),
                resizable=True)

    def _plot_default(self):
        x = np.linspace(0, 2*np.pi, 20)
        y = np.sin(x)
        inv_y = -y
        half_y = 0.5 * y
        inv_half_y = -0.5 * y
        z = np.zeros(x.shape)
        data = ArrayPlotData(x=x, y=y, inv_y=inv_y, half_y=half_y,
                             inv_half_y=inv_half_y, z=z)
        plot = Plot(data)
        plot.plot(("x", "y"), type="scatter", marker="pentagon",
                  marker_size=10)
        plot.plot(("x", "inv_y"), type="scatter", color="blue", marker="hexagon",
                  marker_size=10)
        plot.plot(("x", "half_y"), type="scatter", color="green", marker="left_triangle",
                  marker_size=10)
        plot.plot(("x", "inv_half_y"), type="scatter", color="red", marker="hexagon2",
                  marker_size=10)
        plot.plot(("x", "z"), type="scatter", color="red", marker="star",
                  marker_size=10)
        return plot


MyPlot().configure_traits()

On OSX, it is giving me:

Feedback requested.

[jonathanrocher]

At first sight, the failures seem to have nothing to do with my code so I won't address them in this PR.

[corranwebster]

Thanks for the contributions. Looks like failures are in the svg libraries, so definitely unrelated to this.

Minor criticisms: your hexagons are sufficiently distorted that it bugs me, so you might want to fix. See matplotlib for better ones https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/markers.py#L592. Similarly the pentagons are a bit off in their proportions.

In general for an n-gon rotated by offset radians you want vertices at x = cos(2*pi*i/n + offset)*size and y = sin(2*pi*i/n + offset)*size but for speed you want to hard-code (possibly approximate) values of the sines and the cosines.

Finally, matplotlib's star marker is an actual 5-pointed star: ⭐ I think it makes sense to follow suit, so I'd either remove star from this PR, or follow suit.

Tests would be nice, but I'm not going to insist :)

[jonathanrocher]

Thanks for the feedback @corranwebster . Agreed with all of the comments and I took the time to do the star properly: Updated plot looks like:

[corranwebster]

I don't think that we want this: the markers that have this set have corresponding Kiva versions implemented by Agg in C++ without aliasing, so this is to give pixel-to-pixel similarity between the two ways of drawing between different backends. For these, we have no corresponding Agg versions (unless you want to expose the ones here: https://github.com/enthought/enable/blob/dddfac16132051e40d93485333f1c13ea7964631/kiva/agg/agg-24/include/agg_renderer_markers.h by adding them here:

enable/kiva/agg/src/constants.i

Line 57 in dddfac1

kiva_marker_to_agg = {}

and possibly opengl implementations here:

enable/kiva/agg/src/gl_graphics_context.cpp

Line 63 in dddfac1

glMatrixMode(GL_MODELVIEW);

...)

They will draw slightly slower, but look much nicer if antialias is True.

[corranwebster]

One last comment above. There are native Agg versions of some of these, so if you are feeling enthusiastic you can expose them and they will render much faster, but that could be a separate PR.

The markers look much nicer.

[jonathanrocher]

Thanks for the antialiasing comment. Fixed. Thanks for the pointers into Kiva. Will have to delay diving into Kiva and Agg for the future...

[brycehendrix]

Excuse me for lurking, but I seem to recall trying to get the Agg markers working many years ago, but they didn't work for some reason. It was probably 6 years ago, so I don't recall the exact reason, though I do know I tried :)

[jonathanrocher]

Thanks for your lurking and the information @brycehendrix .