Paul Eggert wrote:
> > What does "otherwise" mean? Either 'configure' finds the system's strstr
> > OK or it finds problems. What is the third alternative?
>
> The third alternative is that 'configure' does not check whether the
> the system's strstr is OK.
Ah, I see: this is the case when the gnulib module 'strstr' is not used.
> Here's the proposal again, amended as per the above:
>
> #if @REPLACE_STRSTR@
> # undef strstr
> # define strstr rpl_strstr
> #endif
> #if @REPLACE_STRSTR@ || ! @HAVE_DECL_STRSTR@
> extern char *strstr (char const *__haystack, char const *__needle);
> #endif
>
> @REPLACE_STRSTR@ is:
>
> 0 if 'configure' checks the system strstr and found it to be OK
> or absent.
>
> 1 if 'configure' checks the system strstr and found that it
> exists but has problems.
>
> _GL_CHECK_PORTABILITY if 'configure' did not check the system
> strstr.
>
> Maintainers can compile with -D_GL_CHECK_PORTABILITY to
> enable portability checking.
OK, I now better understand what you mean. The proposal I'm going for is
basically:
#if @GNULIB_STRSTR@
# if @REPLACE_STRSTR@
# undef strstr
# define strstr rpl_strstr
# endif
# if @REPLACE_STRSTR@ || ! @HAVE_DECL_STRSTR@
extern char *strstr (char const *__haystack, char const *__needle);
# endif
#elif _GL_CHECK_PORTABILITY
# define strstr strstr_in_unportable__use_gnulib_module_strstr
#endif
where @REPLACE_STRSTR@ is undefined in the "otherwise" case.
What's the difference:
1) If the maintainer has not included the strstr module but uses the
strstr function and activates _GL_CHECK_PORTABILITY: With your proposal
he will get a link error about function 'rpl_strstr', and will likely
report it as a bug in gnulib. With mine, he gets a link error about
'strstr_in_unportable__use_gnulib_module_strstr' and knows it's a
problem in his use of gnulib.
2) If the maintainer has not included the strstr module but uses the
strstr function and the installer does _not_ set _GL_CHECK_PORTABILITY:
Your proposal will activate the "extern char *strstr ..." declaration,
while mine won't.
Bruno
