| Issue |
180460
|
| Summary |
Incorrect type deduction in a non-mutable lambda that captures by value
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
lakshayg
|
The C++23 standard has an example showing how types are deduced in a lambda: https://timsong-cpp.github.io/cppwp/n4950/expr.prim.id.unqual#example-1. Relevant parts of the example are reproduced below.
```cpp
void f() {
float x;
[=]() -> decltype((x)) { // lambda returns float const& because this lambda is not mutable and
// x is an lvalue
decltype(x) y1; // y1 has type float
decltype((x)) y2 = y1; // y2 has type float const&
return y2;
};
}
```
This example fails to compile with clang https://godbolt.org/z/zqEsa7T8f.
```
<source>:21:12: error: binding reference of type 'float' to value of type 'const float' drops 'const' qualifier
21 | return y2;
| ^~
```
Clang replaces `decltype((x))` with `float&` in the _trailing-return-type_. However, in the _compound-statement_, `decltype((x))` is replaced with `const float&`. This causes the compilation to fail.
It seems that clang is still looking at the outside `x` and not the `x` in the closure generated by the lambda when evaluating the `decltype` in the _trailing-return-type_.
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
