[Bug c++/92580] New: "if constexpr" not discarding branch

[paultargosz86 at googlemail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92580

Bug ID: 92580
   Summary: "if constexpr" not discarding branch
   Product: gcc
   Version: 9.2.1
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c++
  Assignee: unassigned at gcc dot gnu.org
  Reporter: paultargosz86 at googlemail dot com
  Target Milestone: ---

"if constexpr" uses std::is_same to check type. depending on the type its using
different if-branches, that can have or have not member variables. The
"discarded Branch" tries to use nonexistent membervariables and fails
therefore.

https://godbolt.org/z/jw_BAy


differentiation: 
Bug 85689 is static_assert that is not proposed in "if constexpr" e.g.
wg21.link/p0292

[Bug c++/92580] "if constexpr" not discarding branch

[paultargosz86 at googlemail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92580

--- Comment #1 from Paul Targosz  ---
//examplecode without godbolt

#include 
#include 

struct GW_t {

};


template 
struct IP_t {
constexpr static uint8_t a = A;
constexpr static uint8_t b = B;
constexpr static uint8_t c = C;
constexpr static uint8_t d = D;
};

template
void workaround() {
std::cout << "connecting to static IP " << int{my_config::a} << "." <<
int{my_config::b} << "." << int{my_config::c} << "." << int{my_config::d};
}

int main() {

// * This two lines are the different configurations
 using my_config = GW_t;
// using my_config = IP_t<192,168,0,8>;



if constexpr(std::is_same::value) {
std::cout << "connecting to Gateway";
}
else {
// * first line should work but isn't, second line is the
workaround
 std::cout << "connecting to static IP " << int{my_config::a} << "." <<
int{my_config::b} << "." << int{my_config::c} << "." << int{my_config::d};
// workaround();
}

return 0;
}

[Bug c++/92580] "if constexpr" not discarding branch

[paultargosz86 at googlemail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92580

Paul Targosz  changed:

   What|Removed |Added

 Status|RESOLVED|CLOSED

--- Comment #4 from Paul Targosz  ---
closing issue because you're right