yaber...@ca.ibm.com wrote:
> tar.output is kinda big, and so is the enum to get the SYSINT_BUFSIZE
> value...
True, but SYSINT_BUFSIZE is being declared.
And it's being used on the next line
(line 4602 of tar-1.output).
So why is SYSINT_BUFSIZE "undeclared" later? It's
a global identifier; once that's declared,
it should never become undeclared.
Could be a compiler bug.
What happens if you remove every line starting with "#"
from tar-1.output, put the result into a file called
"foo.c", and compile foo.c? Here's what happens on
my Ubuntu 13.10 host when I run "gcc -c foo.c"
(these warnings may be due to architectural differences).
I don't see a problem with SYSINT_BUFSIZE here.
foo.c: In function ‘append_incremental_renames’:
foo.c:4706:170: warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
( (&stk)->temp.tempint = (char *) (0) - (char *) (&stk)->chunk,
((((&stk)->temp.tempint > 0 && (&stk)->temp.tempint < (&stk)->chunk_limit -
(char *) (&stk)->chunk)) ? (int) ((&stk)->next_free = (&stk)->object_base =
(&stk)->temp.tempint + (char *) (&stk)->chunk) : (((obstack_free) ((&stk),
(&stk)->temp.tempint + (char *) (&stk)->chunk), 0), 0)));
^
foo.c: In function ‘read_incr_db_2’:
foo.c:4942:180: warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
( (&stk)->temp.tempint = (char *) (content) - (char *) (&stk)->chunk,
((((&stk)->temp.tempint > 0 && (&stk)->temp.tempint < (&stk)->chunk_limit -
(char *) (&stk)->chunk)) ? (int) ((&stk)->next_free = (&stk)->object_base =
(&stk)->temp.tempint + (char *) (&stk)->chunk) : (((obstack_free) ((&stk),
(&stk)->temp.tempint + (char *) (&stk)->chunk), 0), 0)));
^
