False positive for style: 'unreadVariable' in some cases when a reference is used

[ramonunch]

I encountered this problem with a code that I simplified a first time to the following (kinda close to what I was actually doing):

void funk(char *str, bool ignorestr)
{
    char buf[32], *p=buf;

    if (ignorestr)
        buf[0] = '\0'; // style: Variable 'buf[0]' is assigned a value that is never used. [unreadVariable]
    else
        p = str;

    puts(p); // but p can point to buf!
}

Then I thought a bit more and simplified even more the problem like this: and with the following code I still get the unreadVariable style warning using the latest 2.11 release

void f(void)
{
    int x[1], *p = x;
    x[0] = 42; // style: Variable 'x[0]' is assigned a value that is never used. [unreadVariable]

    g(p);
}

However when I use a simple int for x I get no error ie:

void f2(void)
{
    int x, *p = &x;
    x = 42;

    g(p);
}

No warnings here...

So I thought here is a problem with implicit casting from int[N] to int* but using int *p=&x[0] does not fix the warning. Of course avoiding the warning is easy and could help clarifying the code (what I did) but it is still a false positive.

[CHR]

Example added here: https://trac.cppcheck.net/ticket/11139

[Sylvain Joubert]

I also encountered a FP using reference. Not sure if it's the exact same pattern though:

#include <nlohmann/json.hpp>

bool check();
bool double_check();

nlohmann::json test()
{
  nlohmann::json data;
  auto& elem = data["elem"]; // style: Variable 'elem' is assigned a value that is never used. [unreadVariable]

  if (check())
  {
    elem = 42;
  }
  else if (double_check())
  {
    elem = 1;
  }

  return data;
}

Notes:
* I could not easily reduce/reproduce this without using nlohmann::json, so I don't know if this is actually a FP or a missing "library" configuration for this type.
* If I change the else if into an else the warning disappears

[CHR]

Thanks for reporting, ticket is here; https://trac.cppcheck.net/ticket/11817