[Bug c++/92338] New: double/float thought (ambiguous) candidate for C-array index. (Non numeric types aren't.)

[mick.pearson at wildblue dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92338

            Bug ID: 92338
           Summary: double/float thought (ambiguous) candidate for C-array
                    index. (Non numeric types aren't.)
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: mick.pearson at wildblue dot net
  Target Milestone: ---

A double (float) conversion is considered candidate for indexing a C-array. A
double (float) isn't allowed to access an array normally, so it's not good that
this makes an int conversion operator "ambiguous".

I tested this with some online GCC compilers... unfortunately none of them give 
the GCC version. I can't conveniently upgrade my GCC.

{{{

#include <stdio.h>

struct Test
{
    operator int(){ return 0; } 
    operator double(){ return 0; }
};

int main()
{
    printf("Hello World");

    int test[2] = {};
    int test2 = test[Test()];

    return 0;
}

}}}

{{{
Error(s):

source_file.cpp: In function ‘int main()’:
source_file.cpp:22:28: error: conversion from ‘Test’ to ‘long int’ is ambiguous
     int test2 = test[Test()];
                            ^
source_file.cpp:14:5: note: candidate: Test::operator double()
     operator double(){ return 0; }
     ^
source_file.cpp:13:5: note: candidate: Test::operator int()
     operator int(){ return 0; } 
     ^
source_file.cpp:22:9: warning: unused variable ‘test2’ [-Wunused-variable]
     int test2 = test[Test()];
         ^
}}}