High,

  concerning HDF5: well, probably we should go there in the long run. Am I
  really the only one with lots of HDF-4 files around? I might even write a 
  napi for HDF-5, but that would be a very low priority project. Completion
  would depend on the speed (or rather slowness) of our mechs getting the 
  instruments going and the necessity to fix the bugs in my other software.

  When I understand the HDF-5 documentation right, HDF5 will eventually
  support HDF-4 files. Which would be the time to switch for us. But I do
  no feel particularly good about changing horses in mid run. We might confuse
  quite a lot of people. Then HDF-5 does not yet support all necessary platforms:
  notable candidates missing are Intel-Win32 and VMS. 

  Then there seems to be a decision which is needed: HDF5 allows to store
  data in a variety of number formats. It also supports converting from them
  to native number formats as well, so do not worry. But I think we should 
  choose a format and nail that down. Preferably a format which is easily
  converted efficiently for most interesting platforms.

                                  Mark Koennecke