Suggestion: Operator `in` for slices

Quirin Schroll qs.il.paperinik at gmail.com
Sat Dec 18 18:29:56 UTC 2021


The operator `in` between `T` and `T[]` is requested a lot of 
times. Most people suggest it should return a `bool` signifying 
if some value of the array on the right-hand side is equal (`==`) 
to the `T` value on the left-hand side. This is the obvious thing 
to do.

I guess it has never been implemented because this information 
(true/false) is so much less than it could be. If D used 1-based 
indexing, `in` could return the index if found or 0 if not found. 
Because 0 == false, one could use `if (auto x in array)`. 
However, D's indexing 0-based, so 0 is a valid index. `in` cannot 
return a plain `size_t`.

My idea is inspired by how AAs `in` operator does not return a 
(near-useless) `bool`, but a value that can be used like a 
`bool`, but provides much more valuable information: a `T*`. It 
would work returning a `T*` for slices' `in` as well, but it is 
much less useful than the index. (One could get the index of 
something from the pointer to it using pointer subtraction, but 
pointer arithmetic not considered `@safe`.) The index practically 
*is* the pointer.

So, can we return an object that is, for almost all intents and 
purposes, usable as an index, but evaluating it via `if` returns 
`true` or `false` depending on whether the value was found or not.

So, my suggestion is that the `in` operator for slices returns a 
value of this type:
```D
struct InResult
{
     size_t index = size_t.max;
     alias index this;
     bool opCast(T : bool)() const { return index != size_t.max; }
}
```

In fact, this is very much a simple optional type. It takes a 
very unlikely valid value of its underlying type and reinterprets 
it as the invalid value.


More information about the Digitalmars-d mailing list