Simple script-based plotting example doesn't display within ipython 1.1.0 notebook with bokeh 0.3

Using bokeh 0.3 with ipython 1.1.0, the initial example in the tutorial, i.e.,

from bokeh.plotting import *
output_notebook()
from numpy import *
x = linspace(-2*pi, 2*pi, 100)
y = cos(x)
output_file("cos.html")
line(x, y, color="red")
scatter(x, y, marker="square", color="blue")

returns something like

<bokeh.objects.Plot at 0x568a190>

but doesn’t display anything when run within the ipython notebook. Running the show() function will cause the plots to be displayed in a separate browser, though.

[damianavila]

If you call output_notebook(), the show function has to display the plot inside the notebook. In the last release, this flow is working without any problems. BTW, since the last release (0.4.2), we had integrated a new bokeh magic with let you activate a --show mode to render the plots inline without calling the show function. For more information about the magic, check this notebook: http://nbviewer.ipython.org/github/ContinuumIO/bokeh/blob/master/extensions/BokehMagic.ipynb?create=1

[pzwang]

The output* functions are mutually exclusive in the sense that they manipulate a global “session”. There is no structural reason why there could not be two of these, but the design of the bokeh.plotting module was to simulate something akin to pylab-style plotting. In fact, the entire object graph constructed by the various plotting functions in bokeh.plotting can be stored into new session objects. In the examples/glyphs/ directory, there are examples that don’t use a session object except at the very end to actually output the plots.