"James Youngman" <[EMAIL PROTECTED]> wrote:
> On Fri, Nov 28, 2008 at 7:19 PM, Jim Meyering <[EMAIL PROTECTED]> wrote:
>
> First, many thanks for looking at this!
>
>
>> +/* Map the dirent.d_type value, DTYPE, to the corresponding stat.st_mode
>> + S_IF* bit and set ST.st_mode, thus clearing all other bits in that
>> field. */
>> +static void
>> +set_stat_type (struct stat *st, unsigned int dtype)
>> +{
>> + mode_t type;
>> + switch (dtype)
>> + {
>> + case DT_BLK:
>> + type = S_IFBLK;
>> + break;
>> + case DT_CHR:
>> + type = S_IFCHR;
>> + break;
>> + case DT_DIR:
>> + type = S_IFDIR;
>> + break;
>> + case DT_FIFO:
>> + type = S_IFIFO;
>> + break;
>> + case DT_LNK:
>> + type = S_IFLNK;
>> + break;
>> + case DT_REG:
>> + type = S_IFREG;
>> + break;
>> + case DT_SOCK:
>> + type = S_IFSOCK;
>> + break;
>> + default:
>> + type = 0;
>> + }
>> + st->st_mode = dtype << s_ifmt_shift_bits ();
>> +}
>
> As far as I can see, the variable type is assigned in the function
> above, but never used. Did you mean to use "type" rather than
> "dtype" in the expression which assigns to st->st_mode?
Whoa! I most certainly did.
I'll fix that right away.
Good catch.
FYI, it passed find's own tests which do exercise those bits,
so I guess I was lucky (unlucky) that the mapping is 1-to-1.
