| Re: reading and importing EOS-HDF [message #66774 is a reply to message #66772] |
Tue, 09 June 2009 06:04   |
Maarten[1]
Messages: 176 Registered: November 2005
|
Senior Member |
|
|
On Jun 9, 2:39 pm, David Fanning <n...@dfanning.com> wrote:
> Maarten writes:
>> HDF-EOS-2 (based on HDF-4) is fully supported in IDL, see the
>> documentation for the eos_pt_*, eos_gd_* and eos_sw_* routines in the
>> online help. These routines are failrly high-level, although the
>> calling order can be somewhat odd (especially when writing data to a
>> file).
>
> Humm. I guess my definition of "high-level" must be different
> from yours.
It is a matter of gradation. The hdf-5 routines are mostly god-awful,
while the hdf-eos2 routines almost make sense. Part of my experience
is with the C-interface, which IDL follows a little to closely for
comfort. In the C-world, the difference is even more remarkable.
> Nevertheless, I find this alphabetical listing of
> routines you *could* use if you knew what the hell you were
> doing (which you don't) to be totally off-putting and
> depressing.
Agreed, there are things you could do with your time that are more
fun. Still, the eos_sw_* interface isn't totally bogus (open file,
open swath, read data, close swath, close file - makes sense to me).
The eos_gd_* interface has too many options for defining the grid, I
hope your files don't use the most exotic ones.
> Since years of IDL experience doesn't seem to
> help get through the morass, I can only imagine what I would
> think if I were approaching it as an IDL novice. "Matlab",
> probably. :-(
Python, especially if you have to deal with hdf-5. PyTables is the
best interface for HDF-5 on any platform I've used.
> P.S. I do find it immensely cheering that things are
> "fully supported" though. ;-)
For HDF-EOS-2: you can at least create files that are compliant,
something that is impossible with HDF-EOS-5, although you can fake
things to fool most software.
Maarten
|
|
|
|