On 05/20/2010 01:23 PM, Daniel Stenberg wrote:
> On Thu, 20 May 2010, Phil Blundell wrote:
>
>> I'm not sure that adding a mechanism which is explicitly for c-ares
>> would be all that wholesome. Is this for your query source address
>> binding thing? If so, you might do better to just add some kind of
>> generic CURLOPT_DNS_QUERY_SOURCE_ADDRESS option and let the various
>> DNS backends implement it as they wish.
>
> I agree.
>
> And for this particular feature, perhaps it makes sense to let
> CURLOPT_INTERFACE bind the resolver's local end as well? I mean if you
> care about where the resolver traffic goes out, don't you also most
> often want the same control and interface for the actual subsequent
> protocol traffic?

Ok, how does this look? It appears to be working as desired,
though I've only tested the DNS_INTERFACE and DNS_SERVERS API
so far.

I know I need to fix up the c-ares indentation to more closely
match what is already in the files...working on that now.

Thanks,
Ben

-- 
Ben Greear <greearb_at_candelatech.com>
Candela Technologies Inc  http://www.candelatech.com

-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html