On Wed May 25, 2005 at 12:21:35 +1000, Peter Miller wrote:
>On Wed, 2005-05-25 at 09:20 +1000, Benno wrote:
>> This will warn about the name of an argument in a function
>> *declaration*, shadowing a global.
>
>I try very hard to make the parameter names in the function prototype
>match the parameter names in the function definition. This makes the
>Doxygen documentation more useful, and there's less to remember.
Agree.
>Plus one more rule: always include the .h with the interface defintion
>in the .c implementing that interface. This lets the compiler detect
>incorrect proptotype vs defintion (i.e. horrendously hard to find bugs).
Agree.
>In the context of the above two paragraphs, the error is not meaningless
>when compiling the function definition. Fixing the function definition
>to use a different name (because of -Wshadow) implies changing the
>function prototype to match. Viola, no more "meaningless" warnings.
Doesn't fix the problem. The implementation of the functions don't
include. the header with the shadowing. A more full example follows:
### foo.h
int foo(int bar);
### foo.c
#include <foo.h>
int foo(int bar) { return bar * 2; }
/* Foo.c compiles fine w/o shadow warning! */
### bar.h
int bar(int x);
### bar.c
#include <bar.h>
int bar(int x) { return x / 2; }
/* bar.c compiles fine w/o shadow warnings! */
### magical_program.c
#include <bar.h>
#include <foo.h>
/* totally different program written by someon else, uses
foo.h and bar.h and now gets -Wshadow errors!!! */
Cheers,
Benno
--
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
