Hi all, Nicholas Earl reported a problem with his simulations -- 400x200x200 -- that resulted in a breakage in the smoothed covering grid, used for ghost zone and vertex centered data generation. In r1679 I applied a fix for this to the yt trunk, changing the rounding method for calculation of the extents of a covering grid from "ceil" to "rint". I investigated his data and what was happening was simply a precision bug causing the ceil to go over, where I believed when I wrote the code it would always be epsilon below the desired value. This should only affect generation of ghost zones for L=0 grids; if any of you happen to run volume renderings that die on getting vertex centered data from L=0 grids, please let me know immediately and we'll try to fix it in a more robust fashion. -Matt