[Date Prev][Date Next][Thread Prev][Thread Next]
- Subject: Re: bug in gcc?
- From: Dave Dodge <dododge@...>
- Date: Wed, 4 Apr 2007 14:39:32 -0400
On Wed, Apr 04, 2007 at 03:25:56PM -0300, Roberto Ierusalimschy wrote:
> > Unlike 0, the result of a comma expression has a type, int in this
> > case, and int cannot be assigned to a pointer in C++, or without
> > warnings in C (test2.cc), not sure if its a bug.
> That makes sense. What surprised me is that the construction does not
> generate even a warning in C.
That's because the type of the expression:
depends what NULL expands to. In C this is implementation-defined.
gcc is probably doing:
#define NULL ((void*)0)
which causes the comma expression to have type "pointer to void",
which can be implicitly converted to your return type.
g++ on the other hand defines NULL to be 0 because C++ requires it.
In that case the comma expression is not a pointer.
Bear in mind that in C, any of these are valid implementations:
#define NULL 0
#define NULL ((int)0)
#define NULL (sizeof(char) * (int)3.14159 - ('3' - '0'))
So the fact that it works when compiled with gcc is just luck. If you
want to make it portable you'll have to cast the NULL to the proper