Trends in programming languages

I just saw a very insightful comment in the 2003 publication of the IFIP work group 2.3 on programming methodology about one of the paper it contains. I cannot attribute it for sure but since Annabelle McIver and Carroll Morgan are the editors, one of them must be at the origin of the comment. It says that in the field of programming languages, one can distinguish two trends. The first, by far the most popular, views a programming language as a formalism to abstract away from the specifics of the machines and the tiny optimizations that might be useful but are tedious to come up with and to maintain. It gives Fortran and Java as examples. Fortran allows the programmer to keep his concentration on other things than the evaluation of mathematical expressions which is quite useful. On the other hand, Java allows the programmer to put the implementation of classes in one place and the latter does not have to think about the specific details of the use of classes (e.g. dynamic binding and memory management). Personnally, I would call those coding languages: they make programming simpler than writing machine code. The second trend views a programming language as a formalism to help reflect on the problem at hand. I give here two of my own examples: Eiffel and B. The fact that they cover analysis, design and implementation is a telling sign that it does not support the view of a program as a sequence of instruction but as a computational solution to a problem. When trying to "sell" languages of the latter category, I often encounter objection stemming from the interlocutor's adherence to the first category and, most of the time, it ticks me off because I have the impression that they are focusing on the wrong aspect of a programming language. Having seen this description of the two trends, I'm thinking of questioning my interlocutor's yardstick instead of getting frustrated. Simon Hudon September 13th 2009 Zürich