The FDA still uses a format for exchanging
data about drug studies which was introduced in 1999. The format doesn't
include metadata so implementers use a separate XML file that describes the
relationships buried in the data. Interestingly, the date stamp on the
formatted file isn't Y2K compliant (although the data within the file is).
Does this matter? Some would say, if it ain't broke, don't fix it. But it seems to me that with the pace of change in our world, this is too complacent an attitude. More and more, you don't know something is broken until it meets a situation it can't handle – by which time it's too late.
I'd say that even the requirement to align a data format with a descriptive XML file is asking for trouble. Self-describing data has a whole, lengthy list of benefits for developers, integrators and users. A twin-file, “other-describing data” approach isn't the same thing at all.
XML used to be widely used in webservices now everything seems to be JSON because its easier for modern programming languages to read (and lighter weight). The world keeps moving!
Posted by: Aaron Wallrich | Wednesday, 28 November 2012 at 03:14 PM