[
Lists Home |
Date Index |
Thread Index
]
Nicolas LEHUEN wrote
> I am sorry but obtaining information about relationships is
> of little use if
> you don't have the corresponding semantics. It is useful for
> lexical-level
> tools, granted, so that you can pretty-print the document, or check
> well-formedness, but that's all.
XML is designed to physically reflect data structure and provides a
standard, simple mechanism for including basic semantic information.
With a csv file all I know is that there are fields with delimiters, some of
which may be descriptive tags. People don't always use commas for
delimiters, they do not always enclose strings in quotes, field names can
appear anywhere in the structure or not at all, one row does not always
equate to one record, multiline records are not always the same number of
lines (you can express a hierarchy with a csv file, it isn't always obvious
to the human eye, but it can be done). There is no guarrantee that physical
structure in any obvious way reflects logical structure.
Never mind what different people mean by "self describing", xml for all its
faults is a big step forwards to anyone who has to write generic, reusable
code that allows users to manipulate structured data from multiple sources.
Believe me, it goes way beyond pretty printing and checking well-formedness.
Agreed, without knowing what tags mean, my user can do very little, but at
least they don't have to get to grips with many weird and wonderful data
structures, and in the best cases they can look at a new message and just
work out what most of it means.
NB, the context I know best is b-to-b messaging, and believe you me, you do
get some really weird variations on "csv" - largely I think because people
so often try to squeeze complex logical structures into completely
inappropriate physical structures.
All the best
Mark
|
