Howdy -
Just got done tracking down a confusing thing that resulted from my misuse of 3dClustSim. I have the blur estimates in a file, so I was supposed to issue a command like:
3dClustSim -fwhmxyz `cat blur_est.1D` ...
but what I actually did was:
3dClustSim -fwhmxyz blur_est.1D ...
This was easy to confuse partly because other AFNI programs take 1D filenames as arguments; and also because 3dClustSim didn't complain that I passed a non-tuple (non-number in fact) as the argument to -fwhmxyz. I finally tracked this down when I looked in the generated .niml file and noticed that the blur estimates were all set to 0. Nothing else about my misuse upset the workflow -- 3dClustSim didn't generate a .mask in this case, but my script calling 3drefit still attached the clustsim results to a dataset, only generating a warning that the mask filename I told it to expect was missing.
It might be nice to either fail violently if -fwhmxyz gets anything other than a triple of floats; or else to try to interpret the arg as a filename; although that could be a pain to parse since 10.29 (for instance) is a valid filename on most POSIX filesystems. So maybe failing violently is better and easier.
Anyway, figured that might save somebody else some pain.