Filling in some library examples

[jwhendy]

Hi Tim,

Want to take a look at these? I kept it to just the bar and histogram plots so you could provide some feedback. The main issue several of these libraries have is no obvious way to generate an image and return that image into the notebook cell.

plotly
You can see some of the saga here, but there's no great way to get an image from plotly.offline. In the meantime I have a fork which lets me pass a save_img parameter. On my local repo, I have it return an IPython.display.Image() object so that the command is clean. You can see this isn't as nice for other libraries.

I broke down and created an account and creds for using py.image.ishow(), but I only have a free account. In just setting this repo up I exceeded my 100 call limit running make as there's 13 calls per run. It'd be nice to cache or something (or find a way to use offline).

bokeh
You can export, but I don't know how you render the plot directly, as it returns a path.

altair
The documentation says there's no way to save programmatically. I created a tmp_plots dir to save to manually and used alt_plot-type.png for the filename.

pygal
We can use render_to_png, but it returns some cairosvg.svg2png() call which I can't find. I get NoneType for var = p.render_to_png(fname).

Let me know if you ideas on a good route forward given the above. It would be nice if these web-based libs could just render in an iframe/be embedded. This is a lot of hoopla to get images when it's definitely not their native way and they'll end up on the web anyway :)

[tdhopper]

Missed your PRs. I'll take a look when I get a chance.

[tdhopper]

I'm going to close this since the Plotly examples I have now work okay and Altair has been rewritten. I'm interested in trying to get the Bokeh examples in if you want to open a PR just with those examples.