Operating Experience with ALGOL 60
Author(s) -
E. W. Dijkstra
Publication year - 1962
Publication title -
the computer journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.319
H-Index - 64
eISSN - 1460-2067
pISSN - 0010-4620
DOI - 10.1093/comjnl/5.2.125
Subject(s) - computer science , group (periodic table) , programming language , mathematics education , psychology , physics , quantum mechanics
with the tide, and advocated separation of data and instructions, were branded as reactionaries. Today we seem to be coming back to the idea of special-purpose features and special stores for different categories of data and instructions. For instance, the Atlas has four different memories for different categories of information, and machines with lockout bits or memory limit registers effectively use these to differentiate between categories of memory. Before we advocate hardware to implement compilers, we should be clear that we are advocating more special-purpose features. The question is whether that is entirely a good thing. I probably think it is. After all, the most .general machine is a Turing machine, and nobody complains about arithmetic and even floating-point facilities. However, I think that we should be clear about the implication of advocating specialpurpose features. Dr. Clippinger: If we built FACT again, we could do it for fewer man-years. I doubt whether the factor is as high as 100. 1 believe that there is an irreducible minimum involved in something as complex as FACT and that we might get the figure down to ten man-years. If we did, it is an interesting exercise to extrapolate in order to estimate the time which might be taken for the tenth edition of FACT! In the process of doing this, we are learning about compilers and the better ways of handling them, and how to get the job done for fewer man-years. But this has no effect on the end result, which is what the programmer is using. That point should be kept clearly in mind. Mr. d'Agapeyeff: I do not believe that compiler readability will lead to special-purpose machines. It is a general-purpose tool although it can have a wide variety of uses for many other things and certainly we have used it for purposes which were never thought of originally. Special-purpose input/output or the use of interpretative or semi-interpretative techniques will require better machine design. The Deputy-Chairman: There is no more time for questions, but if anyone has anything more to say, he will not have much difficulty about re-phrasing his question to make it appropriate to another session. (Laughter.) I am sure that you would not like the session to conclude without thanking the three speakers who have given our conference such a good start. (Applause.) The Conference Adjourned.
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom