Re: [RFC] Add is_list(mixed $value): bool to check for list-like arrays

Hi Nikita Popov and Benjamin Morel, > > I probably brought this up in a previous thread, but I think it's worth > > considering again here, given recent changes to the RFC: > > > > I think it would make more sense to introduce this as `function > > array_is_list(array $array): bool`. That is, a function that only accepts > > arrays in the first place and determines whether the given array is a list. > > > > Your RFC does mention this possibility, but I think the argument it makes > > against it is not particularly strong, especially given the recent rename. > > The argument is that is_array_and_list($array) is shorter than writing out > > is_array($array) && array_is_list($array) -- that's still true, but with > > the new name, it's really not that much shorter anymore. > > > > On the other hand, is_array($array) && array_is_list($array) cleanly > > separates out the two predicates. If we take into account the fact that in > > the vast majority of cases we will know a-priori that the input is an > > array, just not whether it is a list, making the function > > array_is_list($array) is both clearer and more concise. > > > >    function foo(array $array) { > >        assert(array_is_list($array)); // Already know it's an array... > >    } > > > >    if (is_array($value)) { > >        if (array_is_list($value)) { // Already know it's an array... > >            return serialize_as_list($value); > >        } else { > >            return serialize_as_dict($value); > >        } > >    } > > > > Regards, > > Nikita > > I like array_is_list(array $array): bool much better, too. After seeing the discussion about confusion with potential list/vec types, the potential for confusion with ArrayObject/SplDoublyLinkedList, and the likeliness that this will usually be called on values that are already known to be arrays, array_is_list(array $array): bool seems like a better option - it's better to throw a TypeError for objects despite the slight inconvenience of potentially needing to rule out null/false. I plan to update the RFC/implementation soon. Thanks, -Tyson