XML again

[Mark Koennecke]

Dear David Husk,

On Fri, 21 Dec 2001, Husk.David wrote:

> The reason that I'm looking over NeXus is that at NIEHS we are starting to put
> together a project for scientific data archival and storage.  Probably on the
> order of a few TB to start.  This biggest problem I see with the current data is
> finding a way to index and search it.  Especially after it has been stored for a
> few years in an archive.
> 
> XML because it is well defined, readable as text and rapidly becoming a standard
> So the question becomes will NeXus data anytime in the near future be XML based?
> The data could still be mostly binary with an XML header see
> http://www.infoworld.com/articles/hn/xml/01/04/20/010420hnxml.xml.  
> 
> Any thoughts on this?

  Sorry, for the delay in the reply, the festive season.....

  Currently there are no plans for the development of a separate NeXus-API 
  for XML. You are the first one lobbying for this. 

  If you use the new hdf-5 version of the API you obtain essentially hdf-5 
  files. These can be converted with a standard utility coming with the 
  hdf-5 distribution into XML-files. 

  We have a similar problem like yours with keeping track of a large
  number of files. We solved it in the following way:
  - a cron job triggers a progam any 24 hours which scans for new files
    and extracts interesting information (user, title, etc.) from new
    files and writes this information into a mySQL database. 
  - This database is queried through a WWW interface which accesses the 
    database through Java servlets and the JDBC-API.
  If your are interested in this system I can provide you with a more
  detailed description and if the interest gets even larger with my source
  code.  

              A Happy New Year to You!

			Mark Koennecke