This document attempts to gather the source materials for the development of the ALGOL family of languages. It is suggested that it be studied together with the papers by Alan Perlis and Peter Naur in the ALGOL session of the first HOPL conference and the paper by C. H. Lindsey in the ALGOL68 session of the second HOPL conference. Papers on the history of ALGOL in this web site has additional resources.
"A first attempt to automatize programming for computers was made by H. Rutishauser ... The new language follows essentially the ideas given by Rutishauser in his original paper as stated in Section 1: A formula program, that is description of a computing process in this language, consists in 'ergibt' formulae (describing the arithmetic operations to be carried out), and in 'guiding symbols' (e.g., for i = 1 ... or similar verbal statements) describing the structure of the computing process. Ergibt-formulae and guiding-symbols are called 'statements'. [Bauer et al. 1958]
English translation of previous German document from Zűrich, Mainz, Műnchen, Darmstadt (ZMMD-Projekt). In "The Cellar Principle of State Transition and Storage Allocation", Bauer cites this document as having additional authors P. Graeff, P. Lauchli, and M. Paul and gives its date as 9 May 1958.
''This report gives the technical specifications of a programming language proposed by the Ad Hoc Committee on Languages of the Association for Computing Machinery. ...
The objectives of the Ad Hoc Committee in designing the language described herein were to provide a language suitable for:
(1) publication of computing procedures in a concise and widely understood notation,
and
(2) accurate and convenient programming of computing procedures in a language mechanically translatable into machine programs for a variety of machines."
Formal minutes of the daily activities of the conference were kept. Each evening the day's chairman and secretary had the minutes typed for the use at the next day's session. They represent the record of the stages through which the various language parts evolved to achieve their final form. No one chaired individual sessions. No chairman organized the flow of issues discussed; common sense and common purpose kept the meetings productive. [Perlis, HOPL]
"Editors' Note. In the interest of immediate circulation of the results of the ACM—GAMM committee work on an algebraic programming language, this preliminary report is presented. The language described naturally enough represents a compromise—but one based more on differences of taste than on content or fundamental ideas. Even so, it provides a natural and simple medium for the expression of a large class of algorithms. This report has not been thoroughly examined for errors and inconsistencies. It is anticipated that the committee will prepare a more complete description of the language for later publication.
For all scientific purposes reproduction of this report is explicitly permitted without any charge."
"The ALGOL Bulletin, the European medium for disseminating ALGOL intensive matters, was established at a subsequent European conference in Copenhagen in 1959, and the first of its issues appeared that March. Peter Naur of the Danish Regnecentralen was its first editor." [Perlis, HOPL paper, 1981]
Issue 7 (November 1959) collected a number of proposed changes to ALGOL 58.
Abstract: "The value of such a common language for easy and precise intercommunication has been felt since some time. This is demonstrated by the existence of users' associations (SHARE, USE), which however have solved the communication problem only on the basis of their special computer language. It seemed obvious for practical reasons that a common language could not even be a special computer oriented language, favouring more or less one or some of the existing, in their computer language to a considerable amount differing computers. Without doubt a universal computer oriented language (UNCOL) has importance for a number of technical intercommunication problems including the translation process from a common language to computer languages; but it was felt by the Zűrich Conference that a common language for numerical analysis and for the use of computers in scientific computation should be as close as possible to common mathematical notation, the more as this is already a unifying clement. There was no question that the language like FORTRAN should be an operational constructive language (its semantic meaning being a sequence of operations in real time defined in a constructive way), in order to make it possible to execute the language by a computer or to translate it mechanically into a special computers' language. This excludes, of course, all intentional (implicit definition using) parts of mathematical notation. But special attention has been given by the Conference to the possibilities of using free constructive definitions for numerical procedures as a counterpart to the idea of sub-routines and library routines. Thus the wanted language presents itself as an algorithmic language (ALGOL) in the sense of RUTISHAUSER'S early idea, being a special problem-oriented language. It was the task of the conference to standardize the arithmetic notation and to supplement it in order to make it completely operational."
"Only in October 1959, during a second phase of studying Backus's Paris report, did I penetrate through his formal, syntactic description. This made me realize that his notation would be a highly valuable tool for the kind of description that I had in mind." [Naur, HOPL paper, 1981]
Abstract: "This paper considers a universal language designed for use by scientists whose sole concern is to solve their problems rapidly, without regard to the technical difficulties in coding for different types of computer. Such a language must therefore be quick and easy to learn, suitable for the formulation of all scientific problems, and usable with any sufficiently powerful machine. It begins by making a rapid analysis of the results obtained with existing systems of automatic programming-none of which, despite the progress made, have yet escaped from programming conventions imposed by the computer.
It then discusses the conditions which any proposed universal language must fulfil in order to serve this purpose; in particular, clarity of script, flexibility of use and ease of learning are all essential. In the light of these criteria, the authors examine the algorithmic language proposed at the Zurich conference (27 May to 2 June 1958). The universal language now proposed is based on the classification of mathematical entities which was adopted at that conference. Although it retains certain features of the algorithmic system, it embodies some differences in the forms used for transcribing the programme components, and proposes a new method of drafting.
This language is designed to meet the needs of scientists only, and makes no claim to be exhaustive; but its structure is such that it could be incorporated in an 'expanded' language suitable for the drafting of all matter in the field of information processing."
"Automatic programming (AP) is concerned with transforming problem oriented command languages into machine language codes. The solution of a problem generally begins with the statement of an algorithm which must be given in some notation, preferably one which provides full operational and descriptive power and uses an optimum number of symbols. The need to transform this into machine code itself leads to algorithms which should be written in a problem oriented language. This gives a recurrence procedure and implies that a few problem-oriented languages are sufficient." Short items by Perlis, Holt, Huskey, Kobrinski, Poyen, Riesel, and Bauer & Samelson, followed by a discussion.
"The object of this note is to advocate that the IAL language be extended to include two additional notations: conditional expressions and recursive definitions."
"In view of the constructive criticism that evolved and the proposals made, it was decided that another international ALGOL coference should take place. Accordingly the GAMM subcommittee organized a preliminary meeting at Paris in Nov. of 1959, attended by about 50 participants from Western Europe, from which 7 delegates for the final ALGOL conference were selected." [Rutishauser 1967]
The 7 European delegates met in Mainz in December 1959. "This meeting was almost exclusively concerned with problems that led to the structure of blocks and procedures ..." [Naur, HOPL, page 99]
This section is based on Naur's Reference list to his HOPL Paper, a 1970 letter Naur wrote to Donald Knuth, and Thorkil Naur's curation of the Peter Naur papers. The Metadata and PDF links following each item link to Thorkil's original scans at datamuseum.dk. A number of documents were produced leading up to and during this conference. Naur assigned the document numbers in the 1970 letter to Knuth and used them again in the reference list for his HOPL paper. They precede each item in the list below. See Naur's cover letter for additional notes on the documents:
Includes: Suggestions and Corrections to Mainz Report; The Syntactical Structure Proposed by the European Members,;The European Proposal for Procedures; Semantics of Types and Assignment Statements; Input Parameters; and Output Parameters.
This also appears in Section 2.1.2 of: Peter Naur. Computing: A Human Activity. ACM Press, Addison-Wesley Publishing Company, 1992.
"One last thing I'd like to add is that Peter Naur's conduct of the ALGOL Bulletin and his incredible preparation for that Paris meeting in which ALGOL was all written down already in his notebook--he changed it a little bit in accordance with the wishes of the committee, but it was that stuff that really made ALGOL the language that it is, and it wouldn't have even come about, I don't think, had he not done that. [Backus, HOPL Q&A, 1981]
This section uses the same sources and conventions as the previous one.
Corrections to Supplement 2. 3 October 1960. datamuseum.dk: Metadata / PDF
"In this paper some of the more obscure features of the language are considered and their usefulness is discussed."
"NOTE: This Report is published in the Communications of the ACM, in Numerische Mathematik, and in the Computer Journal. Reproduction of this Report for any purpose is explicitly permitted; reference should be made to this issue of the Communications and to the respective issues of Numerische Mathematik and the Computer Journal as the source."
cartesian and union types.
"1.1. The Working Group 2.1 on ALGOL was set up in 1962 (at the time of the revision of ALGOL 60) by the Technical Committee on Programming Languages (TC2) of the International Federation of Information Processing (IFIP), with the intention of working on a successor to ALGOL 60 (then called ALGOL 6X) which was to be more general and comprehensive, yet simpler, and still intended for scientific calculations. (Membership of WG 2.1 is on an individual basis rather than by National Society as for TC2. The Chairman is Prof. van der Poel of the Netherlands. Average attendance is 20, plus observers.)
1.2 Preparation of proposals for Input/0utput (1) and for subsets of ALGOL 60, (2) now completed, occupied most of the first three meetings of WG2.1.
1.3 At the fourth meeting (Sept. 1964) it was decided to defer the more ambitious generalizations under the heading "ALGOL Y", and to keep to short-term improvements for "ALGOL X" (3). Proposals were to be published in the ALGOL Bulletin before consideration by the Working Group.
1.4 By the time of the Princeton meeting in May 1965 (before the New York Congress) extensive practical trials had been made by N. Wirth and H. Weber of Stanford University of a generalization of ALGOL 60 called EULER (4), using syntax directed methods. Progress in the means of language definition was also reported by J. W. de Bakker (5), working with A. van Wijngaarden (6) at the Mathematical Centre in Amsterdam, but his method was not adopted by the group.
It was then decided that any member who wished to do so should incorporate his understanding of the agreements of the Working Group in a draft proposal for ALGOL X which would be considered at the sixth meeting in October 1965.
N. Wirth, G. Seegmüller (Bavarian Academy of Sciences, Munich), and A. van Wijngaarden submitted such drafts, and these were discussed in October, together with comments received by Wirth on his draft circulated earlier to the Group. van Wijngaarden's report was primarily on a system of language definition, illustrated by a possible ALGOL X as vehicle of expression.
The most extensive comment on Wirth's proposal came from C. A. R. Hoare (of Elliott Brothers (London) Ltd.) who proposed an additional feature for "Record Handling" (7), to replace the somewhat unsatisfactory idea of tree-structured data suggested in Princeton.
Wirth and Hoare are publishing (December 1965) an amalgamation of their work, as a separate unofficial document, and so also is van Wijngaarden publishing his.
1.5 The Working Group appointed a subcommittee of four, namely van Wijngaarden, Wirth, Seegmüller and Hoare, to agree among themselves on a proposal for ALGOL X, based upon advice given to them by the rest of the Group at the October meeting, and this will be given final approval at the next meeting of WG.2.1 in April 1966. The first draft would in fact be written by van Wijngaarden using his method of definition.
..."
"Euler caught the attention of the IFIP Working Group that was engaged in planning the future of ALGOL. The language ALGOL 60, designed by and for numerical mathematicians, had a systematic structure and a concise definition that were appreciated by mathematically trained people but lacked compilers and support by industry. To gain acceptance, its range of application had to be widened. The Working Group assumed the task of proposing a successor and soon split into two camps. On one side were the ambitious who wanted to erect another milestone in language design, and, on the other, those who felt that time was pressing and that an adequately extended ALGOL 60 would be a productive endeavor. I belonged to this second party and submitted a proposal that lost the election. Thereafter, the proposal was improved with contributions from Tony Hoare (a member of the same group) and implemented on Stanford University's first IBM 360. The language later became known as ALGOL W and was used in several universities for teaching purposes." [Wirth 1985]
"We regard the current Report on Algorithmic Language ALGOL 68 as the fruit of an effort to apply a methodology for language definition to a newly designed programming language. We regard the effort as an experiment and professional honesty compels us to state that in our considered opinion we judge the experiment to be a failure in both respects."