[PATCH v1] tests: getcwd.test

[Jay Joshi]

On Mon, Mar 21, 2016 at 4:43 AM, Dmitry V. Levin <ldv at altlinux.org> wrote:
> On Sun, Mar 20, 2016 at 11:12:08AM +0530, Jay Joshi wrote:
>> --- a/tests/Makefile.am
>> +++ b/tests/Makefile.am
>> @@ -49,6 +49,7 @@ libtests_a_SOURCES = \
>>       tail_alloc.c \
>>       tests.h \
>>       tprintf.c \
>> +     print_quoted_string.c \
>>       # end of libtests_a_SOURCES
>
> Please keep the list sorted.
>
>> --- /dev/null
>> +++ b/tests/print_quoted_string.c
>> @@ -0,0 +1,61 @@
>> +#include "tests.h"
>> +
>> +#include <stdio.h>
>> +#include <stdlib.h>
>> +
>> +/* Modified from string_quote() from util.c.
>> + * Assumes str is NUL-terminated.
>> + */
>> +
>> +void
>> +print_quoted_string(const char *str)
>> +{
>> +     unsigned int i=0;
>> +     int c;
>
> Shouldn't "c" have type "unsigned char", or, alternatively,
> shouldn't "str" be cast to "const unsigned char *", to avoid sign
> extension?

Indeed.

>
>> +
>> +     while (c = str[i++]) {
>
> Wouldn't it be better without "i" iterator at all?  e.g.
>         while ((c = *(str++))) {
>
>> +                     default:
>> +                             if (c >= ' ' && c <= 0x7e)
>> +                                     putchar(c);
>> +                             else {
>> +                                     printf("\\");
>
> Wouldn't putchar(c) in cases like this look simpler?
>
>> +                                     if (str[i + 1] >= '0' && str[i + 1] <= '9') {
>
> This check is not correct because "i" already points to the next
> character.
>

I shouldn't have made this mistake. Thanks.
Patch is attached. I've assigned character block of 3-bits to other
characters to avoid text-overflow.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: a.patch
Type: text/x-patch
Size: 4809 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20160321/06331ef0/attachment.bin>