Allow definition of color palettes using RGB-tuples

[mrksr]

Seaborn offers an API to create new color palettes on the fly. These APIs return lists of RGB-Tuples:

In[1]: sns.color_palette()
Out[1]: 
[(0.2980392156862745, 0.4470588235294118, 0.6901960784313725),
 (0.3333333333333333, 0.6588235294117647, 0.40784313725490196),
 (0.7686274509803922, 0.3058823529411765, 0.3215686274509804),
 (0.5058823529411764, 0.4470588235294118, 0.6980392156862745),
 (0.8, 0.7254901960784313, 0.4549019607843137),
 (0.39215686274509803, 0.7098039215686275, 0.803921568627451)]

It would be very nice to be able to use them in Holoviews directly:

%%opts Curve (color=hv.Cycle(values=sns.color_palette()))

hv.Overlay([curve1, curve2, curve3])

However, Cycle (and Palette) assume colors to be defined as hex strings (like #FF0000) or names (like green).
It is possible to convert the seaborn palette like so:

hv.Cycle(values=[hv.plotting.bokeh.util.rgb2hex(c) for c in sns.color_palette()])

But this is quite cumbersome. Ideally, Holoviews would just accept RGB tuples anywhere it accepts hex strings.

[philippjfr]

@mrksr I just tried it and it seems to work fine for me even without converting to hex. I did find that declaring the Cycle directly in the output magic doesn't work, e.g.:

%%opts Curve (color=hv.Cycle(values=sns.color_palette('Set1')))

Seems to throw an error, but if you declare it separately:

cycle = hv.Cycle(values=sns.color_palette('Set1'))

%%opts Curve (color=cycle)
hv.Overlay([hv.Curve(np.arange(10)*i) for i in range(3)])

It seems to work fine in both matplotlib and bokeh. Could you confirm?

[philippjfr]

I've got a fix for the parser so that %%opts Curve (color=hv.Cycle(values=sns.color_palette('Set1'))) just works. Will make a PR later today.

[mrksr]

Indeed, defining the cycle separately seems to work fine. Thank you for the fix!

[philippjfr]

Once the parser fixes referenced above are merged you'll be able to define it inline in the magic as well. I'll close this issue now though.