Motivation for extensible languages

Our discussion of the motivation for extensible languages rests on a basic premise, namely that there exist diverse programming language requirements which are becoming continually more diverse, and that it is of critical importance that each user in the spectrum of users be supplied with a language facility appropriate to his problem area. That is, there are, in fact, "scientific" problems, "data processing" problems," information retrieval" problems, "symbol manipulation" problems, "system programming" problems, and so on and each problem area has its particular and peculiar units of data, unit transactions, and preferred ways of writing. For example, the kinds of unit data and unit transactions include integers which are to be added, floating point numbers to be multiplied, "infinite" precision rational numbers to be divided, strings to be concatenated, records into which a field is to be inserted, files to be sorted, matrices to be inverted, equations to be simplified, sets to be unioned, and so on. A part of our premise is that it is not enough to have a language which is formally sufficient to host the particular data and unit transactions some user has in mind. Rather, it is of critical importance that the kinds of data and unit transactions which he wants to think of as primitive be available, effectively as primitives, in his language facility.