Looking back: “The future belongs to C”
I was flipping through back issues of IEEE Computer in CS and found this letter to the editor.
There are many areas where modern languages (such as perl and python) strike a good balance between programmer effort and CPU time. It seems to me that, excepting those areas, this quote is still largely true.
I would, however, be careful of choosing C on the basis of performance alone as doing so could be premature optimization.
To the Editor:
Ted Lewis' reply to Kenneth Pronovici(“A role for language ‘relics’,” Sept. 1996, p 10) was insulting and shallow. The Cobol and Fortran analogies were totally off the mark. Cobol was not intended to live forever. Fortran, recognized as having weaknesses, was rapidly restandardized several times. Yesteryear's high-level languages were intended as experiments to show that efficient machine code could be generated from a procedural algebraic problem representation in a limited RAM environment.
Lewis' “suggested avenues” such as NewtonScript and JFactory are useful only for writing research papers. They remind me of Occam and Linda. Java and Visual Basic are not great leaps forward, but by-products of the popularization of the computer, suitable for shipping apps quickly to Joe Sixpack.
Tools that solve general problems efficiently live forever. We still computer with transistorized binary logic circuits, and we will be doing so 20 years hence. Consequently, backend software will be written in C 20 years hence because C can efficiently map a general class of procedural algebraic problems onto transistorized binary logic circuits.
Lewis should bear in mind that yesterday's solutions were important steps forward, not fodder for cheap shots fired years later.
Bruce David Wilner
Network Security Laboratories
Bethesda, Me.
wilner@dockmaster.ncsc.mil.