[newbie] XML for scientific data

[Guillaume Dargaud]

I'm starting a new project under linux based on one of my old windows projects. There are plenty of flux of data, from acquisition to elaboration to graphics to final models. The current formats of data are binary, various, highly specific, with compatibilty problems and need to be redefined anyway so I was thinking of using xml to code for some arrays of values together with their creation parameters.

I'm only vaguely aware of XML's possibilities but I hope it can make our various exchanges of data between apps easier (some apps being application servers, others being scientific apps like MathLab, gnuplot...). The alternative is to use the HDF format but I've used it and I'm not crazy about it.

I know I need to define a DTD for those dataset. Now the question is: "How do I get started ?". And is there a list of apps that support XML natively or can interface through it easily (like gnuplot or others) ?

If anyone can recommand some XML tutorial or book with scientific data in mind. Or tell me altogether that it's a bad idea.

 

[Guillaume Dargaud]

One question in particular is bothering me.
If I have an application producing a continuous xml output (say into a named pipe or a tcp socket), it looks like it can be read by another app only from the very begining, otherwise that app will miss the first lines containing the reference to the DTD. Is there a way around that ?