In article <20070604194515.GQ26767@canolog.ninthwonder.com>,
Allen Briggs  <briggs@netbsd.org> wrote:
>On Mon, Jun 04, 2007 at 07:26:41PM +0000, Christos Zoulas wrote:
>> > 	fprintf(stderr,
>> >-	    "usage: %s [-ez] [-f file] [-l length] [-p prefix] cmd [args...]\n",
>> >-	    getprogname());
>> >+	    "usage: %s [-ez] [-b buffersize] [-f file] [-l length]\n"
>> >+	    "       %*.s [-p prefix] cmd [args...]\n",
>> >+	    getprogname(), (int) strlen(getprogname()), "");
>> 
>> Why %*.s? we never do this...
>
>Just to line things up nicely.  No other reason.

Typically we hard-code the spaces in... I guess this is nicer.

$ ps -\?
usage:  ps [-acCehjlmrsSTuvwx] [-k key] [-O|o fmt] [-p pid] [-t tty]
           [-M core] [-N system] [-W swap] [-U username]
        ps [-L]

>
>> >+	long long buffersize;
>> 
>> Why make buffersize long long, since malloc can only handle size_t?
>> Or at least check that buffersize < SIZE_T_MAX.
>
>To match the return type of strsuftoll().  I don't have a problem with
>changing it to size_t and restricting the conversion to SIZE_T_MAX
>instead of the somewhat arbitrary 256MB.

I think we should just check that it is < SIZE_T_MAX before we pass
it to malloc.

christos