For university researchers to be able to use the Chilton Atlas, a major requirement was that the programming language they use was available or that the facilities available were so attractive that they were prepared to make a change.
Many of the computers in the university sector had rudimentary high-level languages such as Mercury Autocode. Algol had received much support in the academic community as a suitable elegant language for both teaching and scientific research applications. The Elliott 803 had Algol as its main higher level language. The English Electric KDF9 also had Algol as the main high-level language with two compilers (Whetstone and Kidsgrove). Whetstone had fast compilation and slow execution, while Kidsgrove was the reverse. The main FORTRAN compilation system, Egdon (similar to Hartran), was largely defined by Culham and Winfrith for use on the AEA's KDF9s. It later was used on several of the university KDF9s as well. The Hartran and Egdon Fortran dialect was compatible.
The attractions of Algol were often its major flaws. Having reserved words made it attractive when writing algorithms but tended to mean programs were input using paper tape generated on Flexowriters capable of producing underlined reserved words.
Not having standard input/output functions as part of the language meant the language was independent of the world around it. It also meant that each manufacturer defined their own input/output functions and ensured they were incompatible with those of their competitors.
Defining the Algol program as a monolithic whole rather than a set of independent functions that could be separately compiled and linked ensured that large suites of programs were almost impossible to write and manage.
Ferranti and Manchester University had developed a good Algol compiler for Atlas that accepted programs input on 7-hole Flexowriter paper tape. The program had to be complete. The input/output functions defined were incompatible with those in use by most university Algol users (on the KDF9 and the Elliott 803).
In each case both the input/output facilities provided and the punched form of the program were completely different from the Atlas ALGOL conventions. The Elliott 803 used five-hole paper tape, the KDF9 eight-hole while Atlas used seven-hole! However, the tape reader on Atlas could read all three forms of tape.
The Chilton Atlas solution was to enhance the basic Atlas Algol compiler by effectively embedding it inside an outer block that provided:
The pre-processor eventually handled 15 different dialects of ALGOL (including both French and Danish dialects). The library functions for both the Elliott 803 and KDF9 were added and it was even possible for the KDF9 ALGOL programs to run with Elliott I/O and vice versa if needed. A card-based dialect was added and a pseudo-binary format that encoded each ALGOL basic symbol, identifier and number as a single column on a card together with a symbol table. This significantly decreased the size of a source program and thus speeded up compilation. A program editor was added to allow the pseudo-binary cards to be updated.
Despite the universities using Algol as their main teaching vehicle, its use by university research workers never approached that of Fortran. Early on, when significant use of Atlas was made by RHEL and Harwell, nearly 80% of the Chilton users used either Fortran or Atlas machine code. The Fortran usage never dropped below 50% even when the number of languages available increased. The use of Algol never rose above 20%. The elegance of the ALGOL language never overcame its basic weaknesses: no defined markup, no standard I/O, no defined libraries, no sub-compilation and no relocatable binary loader.
Installation of the Chilton Atlas started in June 1964 and an at risk service started in October 1964. The Algol system developed at Manchester was working on the Chilton Atlas in October 1964, and the Pre-processor before the official hand over of the machine in May 1965.
