Portal de Periódicos da CAPES

Programming languages for writing system programs

1974; Wiley; Volume: 4; Issue: 1 Linguagem: Inglês

10.1002/spe.4380040102

1097-024X

Christian A. Lang,

Logic, programming, and type systems

Programming Languages for Writing System Programs My laboratory recently took the backward step of buying a new computer.It is a popular, fast, modern machine with floating point hardware, disks and magnetic tapes, unfairly called a 'mini' as no new name seems to have been coined for this class of powerful machine evolved from the minis of yesterday.We are delighted with the hardware.Starting software development with our newly delivered pride and joy, however, created a feeling of d6jB vu, for a common problem reappeared.Reappeared, for it is one that, with the expenditure of much effort, we had solved on a previous machine and enjoyed the fruits of our labours for five years.What language are we to use to write system programs for the new machine ?Assembly code and FORTRAN are the manufacturer's only offerings : neither is suitable.Should we design and implement a system writing language ourselves ?This was the tack we took on our previous machine, but that language was machine dependent.Should we look for a suitable language implemented elsewhere for this type of machine ?Should we look for a language with a machine transferable compiler, not yet necessarily implemented on this machine ?It is sad to reflect that in 1974 there is still no widely acceptable, widely implemented high level system writing language that one might reasonably expect to be supplied by a computer manufacturer in addition to FORTRAN.The writing of systems in a high level language, be they operating systems, editors, compilers, interactive design systems, text manipulation systems or whatever, is widely accepted (universally apart from a largish minority of whiz-kid assembly code programmers) and widely practised.Burroughs have been at it for the past ten years; MULTICS has used a dialect of PL/l since the inception of the project; the BCPL compiler is written in BCPL.Although the practice of using a system writing language, typified by these examples, is common, no one language has achieved really widespread use.Most of the efforts of what one might call the professional language designers, as opposed to those who design and implement a language because they need it to get on with their real work, seem to have concentrated in the past on algorithmic languages, that is, those principally intended for doing sums rather than for implementing systems.FORTRAN, COBOL and ALGOL are long established, with international committees sitting to debate standards and developments.More recently, the professionals have produced languages of a much more advanced and comprehensive nature, intended for a wider variety of work.Of the more modern languages that are at different stages of gaining acceptance, PL/l has the might of IBM behind it, while ALGOL 68 has the backing of IFIP, the International Federation of Information Processing Societies.Why does no internationally accepted system writing language exist ?There are plenty to choose from-BCPL, Bliss, Coral, Imp, Malus, Pascal . ..! It is quite surprising when one thinks how many system programs are written and constantly being reinvented.There have been many pockets of system writing language developments.The results range from the machine-dependent language PL360 for IBM360 machines, through languages like SAL, developed for Atlas machines, to the breed of machine-transferable