Re: [PATCH 1/3] Port to Solaris

[Jilles Tjoelker <jilles@stack.nl>]

On Tue, Jan 18, 2011 at 09:15:40PM -0800, Brian Koropoff wrote:
> - Older Solaris does not support %jd (intmax_t) in format
>   strings, but it does support %lld (long long), which is
>   the same size on all architectures it supports.  Do
>   a configure check for the sizes of both and prefer %lld
>   when it is safe to do so.

Pedantically, that's a regression for systems that do support %jd, and I
expect compilers to warn about it. If %jd is available, it must be used
for intmax_t, and not %lld, because that is for long long, not intmax_t.

What you can do is use PRIdMAX from <inttypes.h>, normally defined as
"jd". You can then define this to "lld" or "jd" if it is not defined.
I think this makes the code uglier (just like your change), but oh well.

> - Older Solaris lacks stdint.h, but inttypes.h provides the
>   same types and works on all platforms I've tried dash on,
>   so just use it instead.

<inttypes.h> is defined to be a superset of <stdint.h>, so that is ok.
It is also needed for the PRIdMAX suggestion above.

> [...]
> diff --git a/src/arith_yacc.c b/src/arith_yacc.c
> index 6c5a720..bf21830 100644
> --- a/src/arith_yacc.c
> +++ b/src/arith_yacc.c
> @@ -33,7 +33,6 @@
>   */
>  
>  #include <inttypes.h>
> -#include <stdint.h>
>  #include <stdlib.h>
>  #include "arith_yacc.h"
>  #include "expand.h"

This is useful regardless as the <stdint.h> is redundant. The
<inttypes.h> is already needed here because of imaxdiv().

By the way, I wonder what the advantage of imaxdiv() above separate %
and / is. Compilers can detect the matching between a % b and a / b and
do it in one operation, and any use of imaxdiv() trips gcc's
-Waggregate-return.

-- 
Jilles Tjoelker