[Bug c/41517] Unexpected behaviour of #pragma in statement context

[jan.echternach at de dot gbs.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41517

--- Comment #5 from Jan Echternach  ---
The testcases changed or added in bug 63326 comment 21 were all about #pragma
omp. I don't understand the parser changes well enough to tell if #pragma GCC
will be treated in the same way, i.e. if it will also produce an error after
if(...) or for(...).

If it does, I think it would be a fine resolution and I wouldn't mind if this
bug was closed as a duplicate. From my point of view as a user, a difference
between C and C++ is harmless as long as there's no silent change in semantics.
In other words, it would be no problem for me if the C parser rejected code
that the C++ parser accepted.

[Bug c/41517] Unexpected behaviour of #pragma in statement context

[jan.echternach at de dot gbs.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41517

Jan Echternach  changed:

   What|Removed |Added

 CC||jan.echternach at de dot 
gbs.com

--- Comment #1 from Jan Echternach  ---
I can confirm this bug in gcc 4.9 (Debian 4.9.2-10). A different example which
also produces wrong code (should return 0, but returns 1):

//#include 
int main ()
{
if (0)
#pragma GCC diagnostic ignored "-Wtype-limits"
{
//printf ("bug\n");
return 1;
}
//printf ("ok\n");
return 0;
}

And an example that fails to compile ("error: ‘else’ without a previous ‘if’"):

void foo (void)
{
if (0)
#pragma GCC diagnostic ignored "-Wtype-limits"
{}
else {}
}