Hello Brian,
So far, I have tried to do this using HDF5 data organized two ways: one where all the data for each component is in one file (for example - Bx.h5, By.h5 etc.), the other way involved having the data separated by refinement grids in 32 files per component (so, Bx_file_*.h5 where "*" is from 0 to 31). With both methods I have run into issues when trying to draw vector or streamline plots.
You can make your life a little bit easier (by not requiring the comparison operator) by either outputting all group members into a single file (carpetiohdf5::one_file_per_group = yes) or by merging the Bx,By and Bz files for each file_XXX using hdf5_merge (build via make configname-utils then found in exe/configname/hdf5_merge).
viewer: The expression was invalid: The expression parser encountered an unexpected token:
conn_cmfe(</home/projects/B5/3d_data_201207012347/By.file_* database:MHD_EVOLVE--By>, Carpet AMR-grid)
^^^
Again, the three carets are under "AMR"
You want to surround "Carpet AMR-grid" by angle brackets ie. "<Carpet AMR-grid>". Also you likely want to add some timestepping magic to the file name. Something like this:
conn_cmfe(</home/projects/By.h5[0]id:MHD_EVOLVE--By>, <Carpet AMR-grid>)
where the "[0]id" is actually explained somewhere in the VisIt pdf manual (though I don't remember right now in which section).
Yours, Roland