Assertion !isValueDependent() && "Expression evaluator can't be called on a dependent expression." failed
#136118
Labels
clang:frontend
Language frontend issues, e.g. anything involving "Sema"
concepts
C++20 concepts
confirmed
Verified by a second party
crash-on-valid
regression:16
Regression in 16 release
Comments
|
@llvm/issue-subscribers-clang-frontend Author: None (MacroModel)
```cpp
#include <cstddef>
#include <utility>
#include <type_traits>
template <typename... Fs>
inline consteval bool allow_multi_result_vector() noexcept
{
return[]<::std::size_t... I>(::std::index_sequence<I...>) constexpr noexcept
{
return ((
[]<typename FsCurr>() constexpr noexcept
-> bool
{
// check irreplaceable
if constexpr (requires { requires ::std::same_as<::std::remove_cvref_t<decltype(FsCurr::allow_multi_result_vector)>, bool>; })
{
constexpr bool tallow{ FsCurr::allow_multi_result_vector };
return tallow;
}
else { return false; }
}.template operator() < Fs...[I] > ()) ||
...);
}(::std::make_index_sequence<sizeof...(Fs)>{});
}
struct A struct B int main() |
|
Confirmed: https://godbolt.org/z/f4xYYobKs |
!isValueDependent() && "Expression evaluator can't be called on a dependent expression." failed
|
Reduced by me and C-Reduce: https://godbolt.org/z/4esWerzK5 template <typename... Fs>
consteval int allow_multi_result_vector() {
[] <int... I> {
([]<typename FsCurr> {
requires {
requires __is_same(decltype(FsCurr::allow_multi_result_vector), int);
};
}.template operator()<Fs...[I]>() ||
...);
};
}
template int allow_multi_result_vector<>(); |
|
CC @cor3ntin |
|
Reduced further: template <typename Fs>
int allow_multi_result_vector() {
([]<typename FsCurr> {
requires {
requires __is_same(decltype(FsCurr::allow_multi_result_vector), int);
};
}.template operator()<Fs>());
}
template int allow_multi_result_vector<int>();https://godbolt.org/z/PrjfWY8x1 This is unrelated to pack indexing but rather deferred constraint evaluation: the requires expressions remain untransformed until the final instantiation, at which point we sadly don't have full-level template arguments to go with. |
|
Hmm, seems to be a Clang 16 regression: https://godbolt.org/z/MeYe1ox73 |
|
There are a lot w/ this same assertion but the backtraces are all a little different, this one though seems similar enough to maybe be related: #94811 |
|
git bisect points to 13d44a8 which is a revert unfortunately |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The text was updated successfully, but these errors were encountered: