I added an Introduction.ipynb notebook, see #62.

We might want to improve this to allow JSAnimation as an option instead of the widgets so that people viewing the introduction online can see the time evolution.

Looks good! Moving the stripe plotting code to riemann_utils and incorporating it as a standard option with make_plot_function seems like a good idea.

I wrote a JSAnimation version of the stripe plotting in Shallow_tracer.ipynb. It avoids rasterization of the figures and looks just as nice as the widget version. However, it uses a different approach; it replaces the contents of the plot objects rather than creating a new plot each time. I think this is a preferable approach in general, but it takes a bit more work and it's different from most of our code for the book. The problem here is that the inline matplotlib backend and the JSAnimation code expect different things.

For the links in the introduction, it might be nice to refer to chapters rather then filenames; e.g.

[The chapter on shallow water](Shallow_water.ipynb)

rather than

[Shallow_water.ipynb](Shallow_water.ipynb)

One other issue: personally, I think it would be better if these widgets started with t=0. But for the print book we'll want to use a non-zero time.

@ketch: I changed to t=0 but your modification of snapshot_widgets.py in #69 doesn't reset to t=0.2 because elsewhere t=0 is converted to t=0.001 to avoid a divide by zero at the initial time.

This is rather clumsy but I managed to make a version of interact that creates a JSAnimation instead of the interactive widget. I had to introduce a fig parameter to get this to work properly for all three styles of interact since for JSAnimation it has to accumulate a list of figures.

It is illustrated in Introduction.ipynb.

Just tried the interacts with JSAnimation; they work quite well for me. Nothing clumsy.

What's clumsy is that the function exact_solvers/shallow_water.plot_shallow_water_demo has to take a parameter fig and uses this to determine whether to show the figures or not depending on whether JSAnimation is being used, and then in the Introduction.ipynb notebook fig=ipywidgets.fixed(0) is passed in so that it doesn't make a widget for the parameter fig.

So this jsanimate.interact can't be used in other notebooks currently or it produces plots at all output times, not what we want. Making it work more generally would require modifying a bunch of our other functions that make plot functions and corresponding calls to animate.

For now at least I think it's ok to just have the jsanimate version in the Introduction, and then we can discuss later whether to try to make it work more generally.

@ketch: do you understand why Travis is failing?

I suggest closing this in favor of #71.