Naturally, if this is viewed as too silly, no worries - just say no. But I think we're unlikely to find any other more authoritative source for these colors - a quick search reveals only "sage" as part of palettes that people made up for their own purposes (like branding).

I have nothing fundamentally against this.

Documentation will need to be updated. Currently, the list of available
color names is described as being from the HTML/CSS specifications:
http://www.w3schools.com/html/html_colornames.asp (at least I think it is
documented somewhere, I certainly made it very clear in my tutorial).

You mean http://www.w3.org/TR/css3-color/#svg-color in particular? If that is indeed exactly the same list, then maybe it wouldn't be wise to add to that list unless W3C does. Of course, we could just hack our own copy of mpl or add it to our color module, but that might be a bit kwazy.

I'm not sure why this fails on the Python 2.6 build - only.

I think that failure is unrelated to this PR and can be ignored.

I'm in favour of this color addition, but it is worth noting that there are a further 845 colors in the XKCD colors which are not in matplotlib: http://nbviewer.ipython.org/6900318

I've been meaning to get around to looking at the whole of colors and cm - perhaps this is my way in 😉

Why do all roads lead back to xkcd? 😉

I think this is worth merging as-is. There is no need to hold this up on including the rest of that list.