On Saturday, 1 August 2015 at 17:55:06 UTC, John Colvin wrote:
On Saturday, 1 August 2015 at 09:35:53 UTC, DLearner wrote:
Does the D language set in stone that the first element of an array _has_ to be index zero?

For the builtin slice types? Yes, set in stone.

Wouldn't starting array elements at one avoid the common 'off-by-one' logic error, it does
seem more natural to begin a count at 1.

Actually, maybe even better to allow array definitions of form
int foo[x:y];
(y >= x) creating integer variables foo[x], foo[x+1],...,foo[y].

I think the (very old) IBM PL/I language was like this.

See https://www.cs.utexas.edu/users/EWD/transcriptions/EWD08xx/EWD831.html

As other commenters have/will point out, you can easily define a custom type in D that behaves as you describe, and - Dijkstra notwithstanding - there are valid uses for such things.

D is a C derivative, so it seems a shame not to identify causes of bugs in C,
and design them out in D.

For example, in C difficult to write non-trivial commercial programs without using pointers.
Pointer manipulation has a terrible reputation for bugs.
But in D, easy to write commercial programs without using pointers.
Problem has been designed away.

Similarly, off-by-one array bugs are commonplace in C.
We should seek to eliminate the source of those bugs, which basically reduces to the
issue that programmers find it unnatural to start a count at zero.
Whether they _should_ find a zero start unnatural is irrelevant - they just do as an observed fact, so let's change the language so the issue is avoided (designed away).

Suggestion: if the codebase for D is considered so large that zero-basing cannot now be changed, why not extend the language to allow for array definitions like 'int[x:y] foo'? And then have a rule that 'int[:y] bar' defines a 1-based array of y elements?

Reply via email to