On 4/12/12 12:25 PM, Petr Sumbera wrote:
> Hi,
>
> we are running Bash 3.2.38 on Solaris 10 where asprintf() isn't available.
> Thus code in snprintf.c is used.
>
> There is a problem with following command:
>
> bash -c 'printf "x%10.0fx\n" 123'
> x0000000123x
>
> Where correct output should be:
> x 123x
>
> It seems that '0' after '.' enables zero padding. Please see proposed patch
> bellow:
>
> --- bash-3.2.48/lib/sh/snprintf.c Tue Dec 12 12:10:18 2006
> +++ bash-3.2.48/lib/sh/snprintf.c Thu Apr 12 08:55:44 2012
> @@ -1277,6 +1277,8 @@
> data->flags |= PF_ALTFORM;
> continue;
> case '0':
> + if (data->flags & PF_DOT)
> + continue;
> data->flags |= PF_ZEROPAD;
> data->pad = '0';
> continue;
>
> The same code seems to be also in Bash 4.1 so I guess the problem is still
> there.
>
> Any comments?
Thanks for the report. Try this slightly improved patch; yours (and the
original code) doesn't treat a precision specifier beginning with a `0'
correctly. (And the test has to use `printf -v' to exercise the right
code in bash-4.0 and later.)
Chet
--
``The lyf so short, the craft so long to lerne.'' - Chaucer
``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, ITS, CWRU [email protected] http://cnswww.cns.cwru.edu/~chet/
*** ../bash-4.2-patched/lib/sh/snprintf.c 2010-05-30 18:34:52.000000000 -0400
--- lib/sh/snprintf.c 2012-04-12 20:12:56.000000000 -0400
***************
*** 1296,1303 ****
data->flags |= PF_ALTFORM;
continue;
- case '0':
- data->flags |= PF_ZEROPAD;
- data->pad = '0';
- continue;
case '*':
if (data->flags & PF_DOT)
--- 1289,1292 ----
***************
*** 1330,1333 ****
--- 1319,1329 ----
continue;
+ case '0':
+ if ((data->flags & PF_DOT) == 0)
+ {
+ data->flags |= PF_ZEROPAD;
+ data->pad = '0';
+ continue;
+ }
case '1': case '2': case '3':
case '4': case '5': case '6':
