Logo IMG
HOME > PAST ISSUE > Article Detail



Celebrating 25 years of celebrating computation

Brian Hayes

A Wish List

The dream of Kemeny and Kurtz was "programming for everyone," based on the conviction that getting answers out of a computer should be seen as an essential skill in a technological society. But the standard apparatus of software development is far from ideal for teaching or learning this kind of programming. If computing is to be a broadly practiced art form, we need an improved infrastructure for inquisitive programming. Here's my wish-list for a programming environment:

It should be self-contained, easily installed (and uninstalled), well-documented, internally consistent, bug-free and foolproof. Perhaps it's a scandal that GUI-coddled computer users have forgotten how to install an executable in usr/local/bin and add it to their $PATH, but that's the way the world is.

I don't want just to write programs; I want to have a dialogue with the computer. I need to know the answer to one question before I can decide what to ask next. Thus the typical unit of interaction should be a single statement or expression. This kind of incremental computing can't work if a language requires a lot of scaffolding—header files, a "main" procedure—just to run a fragment of code.

I should be able to think in terms of the problem at hand, not the innards of the machine. I don't want to worry about the encoding of characters or whether binary numbers come big end or little end first. I want to be insulated from annoyances such as allocating and reclaiming memory. Data-type declarations should be optional.

I want a system that's mathematically well-behaved. I should not have to live in a world where the integers end at 4,294,967,295 or where 1/3×3 is equal to something other than 1. Give me unlimited precision and exact arithmetic, at least for rational numbers.

I want a rich selection of ready-made functions and data structures. I shouldn't have to build lists or trees or queues out of lower-level constructs such as pointers.

Pamper me. I want the source-code editor with the leather seats, the walnut and the cupholders.

I want a thriving community of enthusiasts, who contribute to the system, share code and knowledge, and keep us all feeling young. For this to happen, the software needs to be free, or close to it.

Along with the wish list there's an unwish list of things I'm willing to give up. Computational efficiency is one of them. I'll typically spend longer writing (and testing and debugging) a program than I will running it. Another variable I'm willing to leave unspecified is the programming language itself. I fervently believe that some languages are better than others, but the programming environment is more important.

comments powered by Disqus


Subscribe to American Scientist