[concepts] improve diagnostics for range concept failures

[ldalessa]

Bugzilla Link	51918
Version	trunk
OS	All
CC	@zygoloid

Extended Description

Clang produces some concepts diagnostics that are of limited use to the programmer, particularly with respect to the ranges library.

In this case, I was playing around with proxy iterators and ran into https://godbolt.org/z/e9M9T3z5j, which tells me that my iterator type fails the borrowed_iterator_t requirement, however it gives me absolutely no information about why.

The borrowed_iterator_t is not a trivial concept, when I google for it I get https://en.cppreference.com/w/cpp/ranges/borrowed_iterator_t which is split into two cases as:

1) `std::ranges::iterator_t<R>` if `R` models `borrowed_range`, `std::ranges::dangling` otherwise.
2) `std::ranges::subrange<std::ranges::iterator_t<R>>` if `R` models `borrowed_range`, `std::ranges::dangling` otherwise.

As a user I have no idea what to do at this point.

The reduced code I was looking at was

struct Broken {
    struct iterator {
        int i;
        auto operator*() const -> decltype(auto) {
            return std::tie(i);
        }
        auto operator++() -> decltype(auto) {
            ++i;
            return *this;
        }
        friend bool operator==(iterator const&, iterator const&) = default;
        friend auto operator<=>(iterator const&, iterator const&) = default;
    };

    auto begin() const -> iterator { return {}; }
    auto end() const -> iterator { return {}; }
};

int foo() {
    Broken r;
    std::ranges::for_each(r, [](auto t) { 
        auto [i] = t;
        printf("%d\n", i);
    });
}

[ldalessa]

I seem to have forgotten to say that what I was looking for would be some sort of "because" stack of failures. While this can add verbosity it can also add clarity, and if it wouldn't add clarity in this case then perhaps some more sophisticated option is required.

[llvmbot]

@llvm/issue-subscribers-clang-frontend

Author: Luke D'Alessandro (ldalessa)

| | | | --- | --- | | Bugzilla Link | [51918](https://llvm.org/bz51918) | | Version | trunk | | OS | All | | CC | @zygoloid |

Extended Description

Clang produces some concepts diagnostics that are of limited use to the programmer, particularly with respect to the ranges library.

In this case, I was playing around with proxy iterators and ran into https://godbolt.org/z/e9M9T3z5j, which tells me that my iterator type fails the borrowed_iterator_t requirement, however it gives me absolutely no information about why.

The borrowed_iterator_t is not a trivial concept, when I google for it I get https://en.cppreference.com/w/cpp/ranges/borrowed_iterator_t which is split into two cases as:

1) std::ranges::iterator_t<R> if R models borrowed_range, std::ranges::dangling otherwise.
2) std::ranges::subrange<std::ranges::iterator_t<R>> if R models borrowed_range, std::ranges::dangling otherwise.

As a user I have no idea what to do at this point.

The reduced code I was looking at was

struct Broken {
    struct iterator {
        int i;
        auto operator*() const -> decltype(auto) {
            return std::tie(i);
        }
        auto operator++() -> decltype(auto) {
            ++i;
            return *this;
        }
        friend bool operator==(iterator const&, iterator const&) = default;
        friend auto operator<=>(iterator const&, iterator const&) = default;
    };

    auto begin() const -> iterator { return {}; }
    auto end() const -> iterator { return {}; }
};

int foo() {
    Broken r;
    std::ranges::for_each(r, [](auto t) { 
        auto [i] = t;
        printf("%d\n", i);
    });
}

[cor3ntin]

@ldalessa https://godbolt.org/z/oP3nardbW we improved that somewhat