The obvious and simple way to plot spikes from a randomspike or spikegen
element is to simply send the state field of the spike generator to a graph
with a PLOT or PLOTSCALE message.  This is the method used in the example
Scripts/tutorials/tutorial4.g.

Although this method is often acceptable, it isn't ideal, because a spike
lasts for only a single time step.  Therefore, you will have to use the
simulation time step as the time step for plotting (using up a lot of memory
to store points that are mostly zero), or risk missing spikes which fall
between the plotting intervals.  Also, when viewed at a high resolution, the
plotted spikes will be triangles, rising linearly from zero at the previous
time step, and falling linearly to zero at the next time step.

The solution, suggested by Dieter Jaeger a few years ago, is to use a
script_out element to check for a spike at every timestep, and then draw a
vertical spike by calling the ADDPTS action of an xplot element when a spike
occurs.  This is ilustrated by the following script spikeplot.g, which should be included
after the rest of Scripts/tutorials/tutorial3.g, or modified to fit your own
simulations.

This directory also contains spikebeep.g, a simple example of the use of
script_out to produce a "beep" whenever a spike is detected.

A more elegant solution would be to create an extended object based on an
xgraph object.  This could receive a PLOTSPIKE message from the spike
generator to correctly plot the spike, as well as PLOT and PLOTSCALE messages
for plotting continously varying quantities.  (A prize will be given for
for the best implementation submitted to babelnews.)