Benjamin Deonovic

I think instead of `insupport(D,a,b)` he means checking whether `D` is a valid distribution with values `a` and `b`. For example if you had a Poisson distribution and tried giving...

1. If the try/catch block were only around the distribution constructor do we really need to worry about other possible sources for an `ArgumentError`? This seems to be a reasonable...

@brian-j-smith Ah right, I was thinking this would be possible inside of the ABC sampler, but you are right this would have to be in the user specified Model, so...

@fisherc2 If you could provide the chain that would be helpful.

@fisherc2 I was able to plot and draw your code. No errors (just some Gadfly warnings) must be an issue with your setup. What version of julia are you on...

To confirm here are the first 2 files from draw:  

error seems to be with autocor. I will investigate further

Issue is one of your variables (divergent__) is all 0s for all iterations and chains. This caused the autocor function to produce all NaNs which caused ChainSummary to error out....

It may be easier to just write your own sampler rather than putting a lot of sampler logic in the Stochastic/Logical node definitions.

For example see `## User-Defined Samplers` in http://mambajl.readthedocs.io/en/latest/tutorial.html or the example samplers in http://mambajl.readthedocs.io/en/latest/examples/pollution.html