I would just like to back up Francois by saying that I would have found it almost impossible to do the sort of science which we did with the AVO Demo, involving selecting and comparing hundreds of images and spectra, at resolutions from 50 mas to arcmin, without heirachical data access. At least, the alternative would have been to do it the old way, searching through long lists by hand or having to zoom in and wait for images to redisplay. For datacubes, it is the answer to a vsiualisation problem which has been bugging me for years and it could be adapted to multi-epoch or multi-polaization data too. Much quicker to have the meta-data available on multiple scales and for images, catalogues and spectra!
I do think that IDHA is still a bit opticl specific but that's easy to fix...
cheers
a