On Tue, Sep 7, 2021 at 6:46 PM Arthur O'Dwyer via Std-Proposals <std-proposals@lists.isocpp.org> wrote:
On Tue, Sep 7, 2021 at 7:10 PM Cleiton Santoia via Std-Proposals <std-proposals@lists.isocpp.org> wrote:
I'm reading p2425r0 and notice that the lambda declaration 

[ captures ] ( params ) lambda-specifiers requires(optional) { body }

in "(params)" part we may include a structured binding syntax but with implicit "auto"

     cv-auto ref-qualifier(optional) [ identifier-list ] 

This could help solve the "Arbitrary lookahead parsing" problem ?

Sample 1 -----------------------
[][a,b] { return a * b; }  // becomes:
[](auto a, auto b) { return  a * b  ; } 

Sample 2 -----------------------
[] const&& [a, b] { return a; }  // becomes:
[](const auto&& a, const auto&& b) { return a; } 

Sample 3 -----------------------
[]&[p] p.foo();
[](auto& p) noexcept(noexcept(p.foo())) -> decltype(( p.foo()  )) { return  p.foo(); };

Your proposal doesn't have anything to do with structured binding, though. Normally (since C++17), when we see `[a, b]`, it means "we're structured-binding to some structure that is going to be destructured here." This does have applications to function parameters, but the application would be, like,

auto magnitude(auto [x, y]) {
    return sqrt(x*x + y*y);
auto coord = std::pair(3, 4);
auto five = magnitude(coord);

This is not C++20, but it's a fairly obvious notion that has definitely been talked about informally, as well as having users ask about it:

So reusing any part of "structured binding" syntax, anywhere in this general vicinity, for any purpose other than structured binding, is a bad idea IMO.


And formally (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0931r0.pdf) which was rejected in Jacksonville, 5-11.