Eric Wilhelm <ewilhelm@xxxxxxxxxxxxx> writes: > I'm working on an idea to allow programs to assume that the data > will tell them what they need to know instead of having to parse the > data, perform some translation, make an educated guess, and then > think that they know what they need to know. Very interesting! Would you care to expand a bit? From the little I can guess, it seems that, if the data has to tell the program what it needs to know, it has to tell it through a language/protocol that the program understands. Is writing the program so that it understands such language/protocol significantly faster and/or more robust than writing it so that it understands a format?