Re: [PATCH] compat/strcasestr: Include correct header file

[Tomi Ollila <tomi.ollila@iki.fi>]

On Fri, Dec 02 2022, Thomas Schneider wrote:

> As per strcasestr(3) of glibc and FreeBSD, the header that defines
> strcasestr() is string.h, not strings.h.  This may cause compilation,
> and thus detection whether an (optimised) version is available, to
> fail even if the function is available, when implicit declaration and
> pointer conversion do not match.
>
> Signed-off-by: Thomas Schneider <qsx@chaotikum.eu>
> ---
> I discovered this when building with Clang:
>
> qsx@naboo ~/src/notmuch (git)-[tags/0.32.2]
> % gcc -o /dev/null compat/have_strcasestr.c; echo $?
> compat/have_strcasestr.c: In function ‘main’:
> compat/have_strcasestr.c:10:13: warning: implicit declaration of function ‘strcasestr’; did you mean ‘strcasecmp’? [-Wimplicit-function-declaration]
>    10 |     found = strcasestr (haystack, needle);
>       |             ^~~~~~~~~~
>       |             strcasecmp
> compat/have_strcasestr.c:10:11: warning: assignment to ‘char *’ from ‘int’ makes pointer from integer without a cast [-Wint-conversion]
>    10 |     found = strcasestr (haystack, needle);
>       |           ^
> 0
> qsx@naboo ~/src/notmuch (git)-[tags/0.32.2]
> % clang -o /dev/null compat/have_strcasestr.c; echo $?
> compat/have_strcasestr.c:10:13: warning: call to undeclared function 'strcasestr'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
>     found = strcasestr (haystack, needle);
>             ^
> compat/have_strcasestr.c:10:11: error: incompatible integer to pointer conversion assigning to 'char *' from 'int' [-Wint-conversion]
>     found = strcasestr (haystack, needle);
>           ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 1 warning and 1 error generated.
> 1
>
> configure then assumes strcasestr is not available when using Clang, so it
> builds the variant from compat/, which later fails when linking because of
> conflicting symbols.
>
> On a side note, debugging was more complicated that I’m used to, e. g.,
> autoconf’s config.log or similar output.
>
> ---
>  compat/have_strcasestr.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/compat/have_strcasestr.c b/compat/have_strcasestr.c
> index 3cd1838d..d52a62ec 100644
> --- a/compat/have_strcasestr.c
> +++ b/compat/have_strcasestr.c
> @@ -1,5 +1,5 @@
>  #define _GNU_SOURCE
> -#include <strings.h>
> +#include <string.h>

Would it be better to include both strings.h and string.h, in
case some (rare) cases it worked only when strings.h were included
(or would that make the test fail in some cases then...?)


Tomi

>  
>  int
>  main ()
> -- 
> 2.37.4\r