Re: [PATCH] non-null declarations

[Paolo Bonzini]

>> I put __nonnull__ declarations
>>   1) in places where passing a NULL pointer leads to undefined behaviour,
>>   2) in places where POSIX says that the function will always fail when a 
>> NULL
>>      pointer is passed. (For example: tsearch.)
>> glibc uses __nonnull__ declarations only for case 1. But IMO warning also in
>> case 2 will be more good than bad.
>
> Good!  Thanks for doing all that work.

I found this:

+extern int rpl_accept (int, struct sockaddr *, int *) _GL_ARG_NONNULL ((3));

do you plan on adding 2-argument versions of the macro, such as
_GL_ARG_NONNULL2?

>> Two points I'm not sure about:
>>   - Should the macro be called _GL_ARG_NONNULL or GL_ARG_NONNULL?
>>   - Is it worth putting the macro definition (always the same 10 lines of 
>> code)
>>     into a separate file, like done with link-warning.h?
>
> I lean towards _GL_ARG_NONNULL, for slightly less risk of impact
> on application name space.

With OpenGL identifiers starting with GL it's a tough call anyway,
especially for preprocessor identifiers (for others OpenGL uses camel
case).

I'd go for GNULIB_ARG_NONNULL.

Paolo