[BusyBox 0001076]: "ip addr del" does not work

[Tito]

On Thursday 26 October 2006 13:39, Bernhard Fischer wrote:
> On Thu, Oct 26, 2006 at 11:51:13AM +0200, Bernhard Fischer wrote:
> >On Wed, Oct 25, 2006 at 10:01:04PM -0500, Jason Schoon wrote:
> >>On 10/25/06, Denis Vlasenko <vda.linux at googlemail.com> wrote:
> >>>
> >>>On Wednesday 25 October 2006 18:06, Bernhard Fischer wrote:
> >>>> >----------------------------------------------------------------------
> >>>> > Andreas Winter - 10-25-06 03:26
> >>>> >----------------------------------------------------------------------
> >>>> >To be more precise: del should work in addition to delete
> >>>> >
> >>>> >----------------------------------------------------------------------
> >>>> > bernhardf - 10-25-06 09:01
> >>>> >----------------------------------------------------------------------
> >>>> >I disagree. ip addr delete should still be supported, so you cannot
> >>>simply
> >>>> >change "delete" to "del" there.
> >>>> >
> >>>> >The attached snippet uses strncmp(,key,strlen(key)) rather than strcmp.
> >>>> >This does not catch ambiguous args when you compare the string_array.
> >>>> >
> >>>> >For this to work, we'd need to keep track on how many matches we got
> >>>and
> >>>> >bail if matches > 1.
> >>>> >
> >>>> >For the ip applet, the strncmp thing should do it for now, though.
> >>>>
> >>>> What do you think about keeping track of the number of matches to catch
> >>>> if we've seen ambiguous matches in compare_string_array?
> >>>
> >>>int do_ipaddr(int argc, char **argv)
> >>>{
> >>>        static const char *const commands[] = {
> >>>                "add", "delete", "list", "show", "lst", "flush", 0
> >>>        };
> >>>
> >>>I think adding "del" here will be perfectly ok.
> >>>(do_iproute has "del" and "delete, for example).
> 
> nah. Don't like that, it's inaccurate and adds too much bloat.
> >>>
> >>>I am committing this change.
> >>>
> >>>Thank you guys.
> >>>--
> >>>vda
> >>>
> >>
> >>The reason this initially led to more discussion is that there were numerous
> >>short aliases supported in the full version, and possibly other applets as
> >>well.  The thought was to find a way to more generically support them
> >>without needing to increase the size of applets by adding every possible
> >>alias variation as a string.
> >
> >That's why i (still) propose to use strncmp.. :)
> 
> What about adding an "exact_match" param to compare_string_array?
> /* Returns the array number of the string.  */
> int compare_string_array(const char * const string_array[],
> 	const char *key, unsigned exact_match)
> {
>         int i;
> 
>         for (i = 0; string_array[i] != 0; i++) {
> 		if (exact_match) {
> 			if (strncmp(string_array[i], key, strlen(key)) == 0)
> 				return i;
>                 } else {
> 			if (strcmp(string_array[i], key) == 0)
> 				return i;
> 		}
>         }
>         return -i;
> }

Hi, what about this?

static int compare_string_array_internal(const char * const string_array[],
 	const char *key, unsigned exact_match)
 {
         int i;
 
         for (i = 0; string_array[i] != 0; i++) {
 		if (exact_match) {
 			if (strncmp(string_array[i], key,  (exact_match)  ? strlen(key) : strlen(string_array[i])) == 0)
 				return i;
         }
         return -i;
 }

int compare_string_array(const char * const string_array[])
{
		return compare_string_array_internal(string_array, key, 0);
}

int compare_substring_array(const char * const string_array[])
{
		return compare_string_array_internal(string_array, key, 1);
}

> untested and can most likely be written to add only a few bytes.
> Alternatively add a compare_substring_array() that strncmps.

Untested, just a proposal.

Ciao,
Tito