[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: xml as a scientific data format



On Sun, 18 Dec 2005, Russell Shaw wrote:

> Michael Gilbert wrote:
> > Hello all,
> >
> > Does anyone have any good references that make clear and good
> > arguments for representing long sets of scientific (floating point)
> > data in xml files (e.g. strings of base64 encoded data).  I need to
> > make an argument to management for the format over hdf.  My gut tells

votable went this way, but allows for binary inclusions. The latter must
be of a well-documented type, not arbitrary blobs.
  http://www.ivoa.net/Documents/latest/VOT.html

> > me that the transparency and extensibility of xml are the key elements
> > to argue.  Are there any other ideas out there or sites that "sell"
> > xml's strong points in clear language that managment would understand?
> >  Thanks for any thoughts.
>
> IMO, it's best to use xml only for export/import between applications.
> Internal to an application, a faster and simpler custom format should be used.

For many applications this is the most sensible option. However for
"data discovery", (eg rummaging in archives) it is attractive to have
an easily-parsed (and thus indexed) format.

If you think reuse is going to be important for the data sets you are
collating, it might be useful to search the net for instances of hdf5
import/export. My impression is it can be quite complex, which suggests
other formats could be better.

It would be useful to document what decision you came to and why, see
  http://wiki.debian.org/DebianScienceDataFormats

Cheers
Vince



Reply to: