Was ALGOL ever used for “mainstream” programming?

问题:

I know that ALGOL language is super-uber-extremely important as a theoretical language, and it also had a variety of implementations as per Wikipedia.

However, what's unclear is, was ALGOL (pure ALGOL, not any of its derivatives like Simula) ever actually used for any "real" programming in any way?

By "real", I mean used for several good-sized projects other than programming language/CS research, or by a significant number of developers (say, > 1000).

Personally, the only ALGOL programming I have ever done was on paper, thus the curiosity.

回答1:

Algol58 seems to have been the most successful in terms of important applications.

From Wikipedia:

JOVIAL is an acronym for "Jules Own Version of the International Algorithmic Language." The "International Algorithmic Language" was a name originally proposed for ALGOL 58. It was developed to compose software for the electronics of military aircraft by Jules Schwartz in 1959.

Then:

Notable systems using JOVIAL include the Milstar Communications Satellite, Advanced Cruise Missile, B-52, B-1B, B-2 bombers, C-130, C-141, and C-17 transport aircraft, F-111, F-15, F-16 (prior to Block 50), and F-117 fighter aircraft, LANTIRN, U-2 aircraft, E-3 Sentry AWACS aircraft, Navy Aegis cruisers, Army Multiple Launch Rocket System (MLRS), Army UH-60 Black Hawk helicopters, F100, F117, and F119 jet engines, the NORAD air defense & control system (Hughes HME-5118ME system) and RL-10 rocket engines. Airborne radar systems with embedded JOVIAL software include the APG-70, APG-71 and APG-73

ALGOL 68 was used in part of DRA for the same purpose. cf. Wikipedia:

The '''Defence Research Agency''' (normally known as '''DRA'''), was an executive agency of the UK Ministry of Defence (MOD) from April 1991 until April 1995. At the time the DRA was Britain's largest science and technology organisation.

DRA's Algol68 compiler was finally open-sourced in April 1999 and is now available for linux for download from sourceforge. (However an interpreter for "Algol68g" is easier to use).

ICL's Algol68 was/is S3 - It was developed by the UK company International Computers Limited (ICL) for its 2900 Series mainframes. It is a system programming language based on ALGOL 68 but with data types and operators aligned to those offered by the 2900 Series. It was the implementation language of the operating system VME.

There are (at least) two other British operating systems - Flex and Cambridge_CAP_computer - written in Algol68 variants. And also 1 Soviet OS: Эльбрус-1 (Elbrus-1), but I have yet to find any of their source code. (If anyone can find and distribute to this source code please let me know)

BTW: I believe that VME is still running - in production - as a Linux/Unixware guest VM. Mostly at Commonwealth of Nations Custom/Immigration services.

Also over the same period the USSR was using Algol68, c.f. history link. Algol68 is used in Russian telephone exchanges. And Algol58 was used in the Russian "Buran/Буран" Space Shuttle landing system.

ALGOL68 was internationalized in 1968. I suspect there are other Algol projects in other countries. esp in German, in Dutch Japanese and Chinese but I have no details.

If you want to actually tryout Algol68 and/or contribute your code, check out Rosettacode's ALGOL 68 repository, then as a class project try one of the "Tasks not implemented".

回答2:

Nothing like responding to 2 year old threads. I program in ALGOL almost daily. I am a programmer on a Unisys ClearPath mainframe and the majority of the system code is written in ALGOL or variants. The Burroughs B5500 was really designed around the language so it is a pretty efficient language/compilation process. Granted, this version is ALGOL with some extensions like limited classes (structure blocks), etc.

i := 80;
while i > 0 do
   begin
   scan ptrRay:ptrRay for i:i until in ALPHA;
   scan ptrEnd:ptrRay for i:i while in ALPHA;
   if i > 0 then
      begin
      replace nextToken by ptrRay for (offset(ptrEnd) - offset(ptrRay)); 
      end;
   end;

That code scans for ALPHA only tokens. It uses the OFFSET function which is a little more costly than using the residual count math yourself (i, starti, etc);

回答3:

Like Tom, I program in ALGOL almost daily - and I'm also on a Unisys Clearpath. ALGOL has been the primary source of my mortgage repayments for more years than I care to remember.

回答4:

When I started programming, Algol was the only compiler available. Yes, it was mainstream till we got a Fortran compiler.

回答5:

To follow up on themis' answer, the entire Burroughs "large system" family (5000, 5500, 5700, 6500, 6700...) was really designed to run Algol well. The operating system, compilers, and major system utilities were written in Algol; if that's not "real" programming, what is?

To be precise, over the life of the product family Burroughs extended Algol into a superset called ESPOL. When Burroughs brought out the "small systems" family (1700, 1800, 1900 series), they defined another Algol-like language called SDL (Systems Development Language) in which the operating software of that line was written. Access to SDL was restricted for security reasons. A variant of SDL was subsequently created with a few of the "priveleged" features removed. The resulting language, called UPL (User Programming Language), was available for customer use.

Some of us still remember when the phrase "Algol-like language" was used to describe any programming language with block-oriented control structures and variable scoping. Widely-known examples of Algol-like languages included PL/I, Pascal, and (...wait for it...) C.

回答6:

Algol was the major programming language for the Burroughs B5000.

回答7:

However, what's unclear is, was Algol (pure Algol, not any of its derivatives like Simula) ever actually used for any "real" programming in any way?

Please, avoid the term "real" programming. "Real" - as opposed to what ? Imaginative ?

By "real", I mean used for several good-sized projects other than programming language/CS research, or by a significant number of developers (say, > 1000).

Yes. It was used for a certain number of projects on which worked a certain number of developers.

Only, what is usually misinterpreted often today is this; in those days computers weren't exactly a household commodity. Hell, they weren't that 30 years ago, less alone 60.
Programming was done in computer centres which were either in goverment ownership (military, academic, institutes of various kinds) or in private enterprises (large companies). And programming wasn't a profession - it was something which engineers, mathematicians, scientiscs and the like used to do when their work was done on paper ... or they had specialized operators which did it for them. Often women, who may or may have not had a scientific background in that particular field - they were "language translators", in lack of a better term (and my bad english).

Programming theories and research was at its beginnings ... vendors being few (and naturally uncooperative to each other) ... each of them used their own extensions, and often programs written for one didn't work well with the other vendor's systems.
There wasn't a "right way" to do something ... you had that and that, and you used whatever catch you could figure to work around your problem.

But, I've wandered off. Let me get back to the number of people. This also goes for several other languages; fortran and cobol, for example. People say, "very few use it". That's simply not true. What is true is that a small percentage of people uses it today, but a larger percent of people used to use it.

As I said, in those days only the sci. and eng. community used to do it. And their number was relatively small, compared to the total population. Nowadays, everybody uses computers, but the absolute number of engineers, mathematicians and the like, is pretty much the same. So it seems that nobody uses those languages anymore ... while in reality, for certain specialized languages (well, nowadays this goes for fortran and cobol, more than algol) the number of users is pretty much constant.

Personally, the only Algol programming I have ever done was on paper, thus the curiosity.

I know I didn't answer your question, but just wanted to clear this. Algol was a little "beofre my time".

回答8:

My first programming experience was on a Burroughs B5500 owned by Northern Natural Gas Company starting in 1970. I started out in COBOL but switched to ALGOL (actually used both) when they needed additional support for a large Oil & Gas Lease Information system that was written almost entirely in ALGOL. At the time there were two programming departments, Business Systems and Scientific Computing. The Scientific Computing department programmed in ALGOL and FORTRAN while the Business Systems department was mostly COBOL with a smattering of ALGOL. Northern advanced from the B5500 to B6500, B6700, B6900, B7800, and B7900 while I was there. I eventually transferred to the Technical Support department and got into making and supporting MCP patches to customize the system for Northern's needs. That was fun!

Short answer to the question. Yes. Northern had a number of application systems written in ALGOL. Of course it was Burrough's version of ALGOL (extended ALGOL).

回答9:

Burroughs B5500 Extended Algol was used heavily for research in astrophysics, linguistics, and statistics at my university (Monash University, Australia) in the late 60s. It was also used in commercial applications that helped pay the bills for the computer center.

As I write this I am running Algol programs in the latest release of the Burroughs B5500 emulator from the team at retro-b5500 in Tasmania. The emulator runs entirely in the browser and faithfully models the processors, disks, tapes, card readers, line printers, card punches and datacom gear!

You can read about the project at http://retro-b5500.blogspot.com/ and http://code.google.com/p/retro-b5500 and you can write Algol programs for arguably the finest Algol machine ever made (except perhaps its successor the B6700.)

One of the postdocs from Monash wrote a reverse compiler from IBM Assembler to Burroughs COBOL in Algol, which was used to migrate all the billing applications at the state-run Gas & Fuel Corporation from IBM 360s to Burroughs 6700s.

回答10:

Back in 1970, I helped develop a Jovial compiler for the Royal Dutch Navy. One of its big advantages was that it was written in Jovial, hence we all got to become pretty good Jovial experts. In fact, as part of the test cycle we would compile the compiler though the latest incarnation of itself and run all our test sets on that. If it passed we would release the first compiler. Thus each release had the capability of compiling itself and that compiler could pass all tests. As every found bug was always added to our self-checking test set the quality of the compiler improved and improved. By the time we left the project we had no known bugs...my once and only time that ever happened.

回答11:

I programmed in Algol/Jovial back in the 70's for the military. I loved the language. You couldn't do recursion in Fortran and I often could make a program much easier by using the correct data structure and a little recursion.

After I had left that assignment, I found that the other developers on the project didn't want to maintain the Jovial code and tried to replicate what I had done in Fortran. It just didn't work and was much slower.

I learned about compiler theory by digging into the source code for the Jovial compiler. Ah... those were the days.

回答12:

Algol was well implemented on the Elliott 4100 machine and was used extensively to develop early refinery process simulations at BP Research center in the late 60s. However, at that time Input/output was not well defined (varied between machines) and at BP it was quickly overtaken by Fortran IV as programs written in strict Fortran IV would run on almost any machine variation - IBM, Univac, Atlas, etc., etc.