Plotting with matplotlib looks blocky in notebook

[Chris Richardson]

At the moment, the behaviour of plot is like this:

mesh = UnitSquareMesh(4,4)
Q = FunctionSpace(mesh, "CG", 1)
F = Function(Q)
F.interpolate(Expression("cos(x[0])*sin(x[1])", degree=1))
plot(F)

or

plot(F, shading='gouraud')

or

plot(F, mode='contour')

I suggest changing the default to use plt.tricontourf(), with, say, 40 contour levels, to get:

[Prof Garth Wells]

Agree, but let's target the 2016.2 release.

Do we have an option for the number of levels?

[Chris Richardson]

I think it gets passed through with the **kwargs as levels

[Chris Richardson]

It is an easy fix. I can push it today.

[Jan Blechta]

I remind that me or Chris talked to matplotlib developer Jens Nielsen asking for linear shading of triangulations for, I think, tripcolor, which would be optimal for plotting P1 functions. But the result was not very helpful as I remember.

[Prof Garth Wells]

X3DOM is probably the way to go for contour plots.

[Chris Richardson]

Well, I've pushed a branch. See what you think. I think it is better.

[Prof Garth Wells]

I think we can merge this.

[Chris Richardson]

I've got a branch for this - it can be merged now.

[Prof Garth Wells]

Fixed in def1637.