disabling unary "-" for unsigned types

[bearophile]

Walter Bright:

>All the various solutions tried over the decades have failed to catch on. This leads me to believe that either these obvious solutions do not work, or they cause more problems than they fix.<

I was not asking to turn D into Ada, but for a tidier number system. The things I was asking are more or less all implemented in the C# language, that is a successful (and pleasant to use) language.
If you use C# for few days for real, you can get some practice in such things too. If you don't like to install dotnet, you can install mono.


>That's a valid argument if you're writing a spreadsheet program. But programmers should be amiliar with computers, and most definitely should be familiar with 2's complement arithmetic. Similarly, if you do much with floating point, you should be familiar with "What Every Computer Scientist Should Know About Floating-Point Arithmetic"<

I agree that a programmer of a system language has to know how fixnums work and has to know some of the basic things about floating point numbers (in truth I don't think 99.9% of programmers know everything written in that article about FP, it's a long and complex paper).

But the number of integral-related bugs in C/C++ programs show that even expert programmers (that surely know what C fixnums are) when they program sometimes "forget" the finite nature of numbers, probably because a human can keep in mind only a very limited number of things at the same time (and most of those things are other details of the algorithm/problem to implement/solve). So a good language doesn't force the programmer to remember the low level details all the time, it allows the programmer to use and remember those low level details when they are needed, or in high-performance situations.

If you pull the trigger of a revolver it will shoot (if it's loaded, etc), but even expert gun users think a the safety is handy, because it helps you to not keep attention all the time to avoid pulling the trigger. Knowing the nature of binary numbers is necessary, and it's bad to hide their nature in a system language, but in most parts of a program, even written in a system language, I want a safety. There are situations where you want a modulo arithmetic behaviour, but in most situations that's not what you desire (and the number of such bugs in programs shows that the brain of the programmers slip back in the 'school arithmetic mode' often).

(Later I have also said that I'd like a way to help the compiler tell apart the two ways of using numbers, the modulo arithmetic from the finite+overflow one).

Bye,
bearophile