Mailüfterl(ID:5255/mai001)Autocode for Mailüfterl"elegant" Autocode for Mailüfterl (Mailüfterl = spring breeze, as opposed to Whirlwind) People: Related languages
References: in ECIP55 Fachtagung 'Elektronische Rechenmaschinen Und Informationsverarbeitung' [Electronic Digital Computers And Information Processing] Darmstadt, Germany, October 25-27, 1955. Braunschweig, F. Vieweg, 1956. view details in ECIP55 Fachtagung 'Elektronische Rechenmaschinen Und Informationsverarbeitung' [Electronic Digital Computers And Information Processing] Darmstadt, Germany, October 25-27, 1955. Braunschweig, F. Vieweg, 1956. view details in ACM Computing Reviews 3(04) July-August 1962 view details So first of all, let me explain to you the direction we wanted to go in the writing of an ALGOL compiler. We joined the ALGOL enterprise at a time before ALGOL 60 was finished. Actually, the first German publication on ALGOL, on ALGOL 58 as it was called afterwards, is written by me.38 I knew already the from my visits the different people important in Germany in the field of programming languages, and we discussed what German expressions would have to stay for the English terms in German papers. I got a consensus from the German-speaking community, and I published this in the form of a paper.38 Bauer had promised us that we'd get documents which make it trivial to write a compiler -- which, of course, was a wrong assumption. And fortunate, I would say, it was too, because by this way, my next guy, the programmer of the team, Mr. Lucas, really had to dig deep into the art of programming languages and compiler writing. He developed special principles and we ended up by having not only an ALGOL compiler, but also a (smaller) LOGALGOL compiler.39 Because of the reason I will explain immediately, we wanted to have the MAILUFTERL working on logical problems. So we designed a kind of extension of ALGOL for bit chains, for sequences of yes-no decisions as such. We called it LOGALGOL. The key issue is the following. In normal ALGOL operation, in order to store a yes-no decision, say a condition or something like that, any logical value, one has to use a full storage cell for one bit -- no other way. It's treated like the numerical values, zero and one. Working on switching circuitry minimization, one can't afford to do this, because then a chain of 16 bits requires 16 addresses. What is wanted is to make use of the full word capacity for bits. So we decided to have a language with a transition operator which transforms a bit sequence into an entire decimal number and stores it that way. So one can go either way. make all your logical decisions in the processes bit by bit. But in the storage, you have already transformed it, say over the core store, in fully used storage values. And therefore, you can house all the bits you require, which is about, I forget now, but it should be some 500,000 bits, except the program, of course, which you have also to maintain. You can store in all the remainder your operational bits for the switching problem, which we did. Minimization of Boolean functions was, in fact, an important field of application of the computer. Mr. Kudielka has written a doctoral thesis applying a particular algorithm twice on minimization. He had been the guy working on the drum, but then he went into switching. The thesis is based on the Quine principle,40 but it goes quite a step further. That is detail I'm not going into at this time. We also programmed all the cybernetic models in order to show what switching really means in the sense of other application on the computer.40 But the first giant application program was in the field of musical theory. I had the visit of a Viennese twelve sound composer who had a theoretical problem, rather for knowledge purposes than for real composing. A twelve sound series can either be constructed by using each key of an octave once, or else one can do the same by using each interval once. It can be shown that both construction principles make sense. His question was, "How many series exist which fulfill both conditions?" In other words, what was to be computed were all sequences of the twelve sounds where each interval occurs only once. The question is, how many are there and what do they look like? He came already with a kind of program, not an elaborated program, but a logical idea for it. He had also computed one block. We transformed his idea into a real Mailufterl program, and we started by doing the first block. To our satisfaction, there were, like on his sheet, thirteen results. Only, it turned out after the first glory that there were mistakes in it. Of course, it was not machine mistakes. Doing it by hand, he was bound to make mistakes: he had one series twice and didn't see it. And he lacked one and he didn't see it. So, in a run of 60 hours, the machine established the number of such series as two times 1928. That was published41 and since then it belongs to the body of knowledge of the twelve sound composers. The second practical project was a real computational one. My friends of the high frequency department had accepted a project with the question, "How do you feed high frequency energy into a power transmission line, (in a frequency range which happened to be the same as the one which the commercial air traffic is using for the board-to-board communication so that a minimum of the energy sent radiates away?)" They had, in other words, an interest that the power line would not work too much as an antenna. The question was, "Can we adapt the input into the line in such a way that the minimum radiation goes off the line?" They had hoped they would come out with a formula describing the solution, but as it was the trend of the time it turned out only numerical calculation could give the answer. So they were very, very happy that the computer was around. We delivered all the tables they needed, by the computer in a thirty hours program. This is the bulk of the early application. We then went deeply into languages. We moved the computer into the IBM Laboratory. We did numbers of computations for IBM and studies of compiling and of looking into languages, and so on. Abstract: Zemanek, an Austrian computer scientist, begins by describing his early life in Vienna, Austria and experiences in Nazi-occupied Austria. He discusses his engineering education and work in radar technology during World War II. Zemanek then focuses on the development of computers in Austria. Topics include: magnetic drums and magnetic memory, the MAILUFTERL computer (which Zemanek designed and built), the LOGALGOL and other compilers, the University of Vienna where Zemanek worked on his computer, the subsequent sponsorship of the project by International Business Machines Europe, and ALGOL and PL/I language standards development. The interview concludes with Zemanek offering a brief overview of the computer industry in Europe from the end of World War II to the 1980 in ACM Computing Reviews 3(04) July-August 1962 view details Resources
|