Re: Patch to change warn_write_strings to flag_const_strings

[Carlo Wood <carlo at runaway dot xs4all dot nl>]

| For C++, string constants are `const char[]' by default, so I'd like to
| change the flag that controls this behavior in c-common.c, and in any case
| it seems odd to control semantics with a warn_* flag.  What do you think?

Unfortunately, I don't understand the patch :/.
Does this change the behaviour for C++ such that string constants
are ALWAYS const?  Independent of any -W... flag?

Ie,

  void foo(char *s);
  void foo(const char *s);

  ...

  foo("Hello world");

calls _always_ the second foo() ?

I wish it would :)

Thanks,

-- 
 Carlo Wood  <carlo@runaway.xs4all.nl>

PS Maybe I need more sleep, but isn't the following incorrect?
   (behaviour of egcs-2.91.53):

~/tmp>g++ -W -Wall -Wwrite-strings str.cc
str.cc: In function `int main()':
str.cc:15: warning: initialization to `char *' from `const char *' discards const
str.cc:15: warning: unused variable `char * x'
~/tmp>a.out
foo(char *);  "Hello world"		<-- should call foo(const char *);
foo(const char *);  "jaja"
foo(const char *);  "12345"

----------------str.cc:
#include <iostream>

void foo(char *s)
{
  cout << "foo(char *);" << "  \"" << s << '"' << endl;
}

void foo(const char *s)
{
  cout << "foo(const char *);" << "  \"" << s << '"' << endl;
}

main()
{
  char *x = "This must cause a warning";	// <-- str.cc:15
  foo("Hello world");

  const char *p = "jaja";
  foo(p);

  const char q[14] = { '1', '2', '3', '4', '5', 0 };
  foo(q);
}
---------------------------------------------------------------