|John Backus - the Father of Fortran|
|Written by Historian|
|Thursday, 22 November 2018|
Page 3 of 3
In 1954 the team led by Backus published its first paper:
`Preliminary Report , Specifications for the IBM Mathematical FORmula TRANslating System - FORTRAN'.
Notice the IBM in the title. FORTRAN was very much an IBM product and as it happened the success of IBM and FORTRAN went hand in hand.
However the six months turned out to be rather longer - two years passed before the first compiler was available (1956) and it took until April 1957 before working compilers were distributed to customers. It was estimated that the project took 18 programmer years to create.
It consisted of 25,000 lines of machine code on a magnetic tape and it was distributed, complete with bugs, to every 704 installation. So, in the tradition of software projects, it was late but most of the onlookers thought it would never be completed and in retrospect two to three years for the first ever compiler isn't bad. Also in the tradition of software projects night work was the norm. They slept in rented rooms in the Langdon Hotel to use the computer at night when it was free!
Doesn't this sound a familiar story?
The IBM 704 - first with Fortran
Despite the bugs the language was a success and of course we all know the rest. FORTRAN introduced many new ideas - arithmetic assignment, comments, the DO loop, subroutines and functions, formatted input/output - to name just a few. It also seeded the idea of machine independence because a machine that had a FORTRAN compiler could run any FORTRAN program.
Also don't think that this was anything like the language tools you might use today. Of course programs were submitted on paper tape or more likely punched cards. Output was to line printer i.e. paper output. If the program ran into a problem it didn't print an error message - it just stopped. The operator, machines had operators back then, would read the instruction counter from the console lights, this is what they were for, and write it down. The programmer then consulted a "stop book" that listed program positions and what caused the stop usually in terms of the internal workings of the compiler.
The first version of FORTRAN contained many concessions to the 704's hardware - the three-way arithmetic IF branch for example - but these were soon lost in improved versions of the language.
It was a language aimed at scientific computing and it lacked many features needed for other types of computing - no character handling for example. This didn't matter too much because with its success came other languages that specialized in other fields - COBOL for business and ALGOL for academic computing.
With the success of Fortran came the explosion of languages but even today Fortran is still going strong and is still the computer language of science and engineering. FORTRAN has achieved this longevity by constantly adding and transforming itself into something better. It has been said that whatever language is used in the future it will be called Fortran!
Perhaps most importantly the success of Fortran focused everyone's attention on software and made us think of hardware as just a way of running it faster or slower.
As well as FORTRAN Backus also gave use a way of describing the grammar of computer languages called Backus Normal Form (BNF). Oddly FORTRAN was the least suited of languages to be described by BNF but it was used to defined ALGOL and it is still used today for all recursive languages.
In 1960 he worked at the IBM Research Division on a particularly ambiguous project to create a mathematical description of set structure. However this came to nothing and he moved on to consider pure programming again. This time he pioneered what we now call the functional approach to programming. The FP group of languages have never proved as popular as FORTRAN but they have been influential in the development of other functional languages.
Should we call Backus the father of FORTRAN?
He himself gave much of the credit to his co-workers for inventing and writing particular parts of the compiler but it is clear that Backus was the man who made it all happen. His legacy is still with us and probably will be for many more years.
How many hardware men can claim that their machines are still in use after sixty years - software is so much more long lived...
And on that note: it seems that the source code for the original Fortran compiler is missing. We have the source code for Fortran II but not the original.
Anyone know where there might be a copy?
or email your comment to: email@example.com
|Last Updated ( Thursday, 22 November 2018 )|