The true meaning of type

I missed something in my previous post on the many meanings of type. Despite the profusion of seemingly incompatible senses, there is one that is consistent across languages. In every language I know of, type is the difference between an integer and a string.

Some languages, like Bliss and Forth, have no such distinction. Everyone agrees in calling them untyped. Of course they still handle integers and strings, but the language doesn't know which is which. They're distingushed only by how they're used (hence the alternate name operator-typed).

Most languages do distinguish, of course, but they do it in completely different ways. In each language, the mechanism it uses is called type. And this is where the confusion starts. In each language community, the word type becomes attached to the mechanism that language uses, and then to other uses of that mechanism - even those that have nothing to do with distinguishing integers from strings. Eventually users come to believe their sense is the real one, and that all the others must be bastardizations of it. And if they realize how different the senses are, they are tempted to think that they have nothing in common, and that type means nothing at all.

But all these senses reflect different answers to the same underlying problem. There is a common meaning to type across languages, and it is not any of the languages' senses. It is the ill-defined problem of having multiple types of data in one language: integers, floats, functions, arrays, lists, hashtables, strings.

1 comment:

  1. That's it! And that is what most people, esp. in CS, completely. Why are types useful? why can they be useful? why can we have them at all? "Type" means something outside prog & CS, namely ~ category, kind, sort; here type of information (--> data).


It's OK to comment on old posts.