RFC compliant address selection vs. home made getaddrinfo

Matthew Hall mhall at mhcomputing.net
Wed Jun 12 15:17:12 PDT 2013


On Thu, Jun 13, 2013 at 12:11:34AM +0200, Bernd Eckenfels wrote:
> dns(only)-based providers are not the best option for general
> purpose resolving because (besides the fact that they will not use
> alternative naming methods which are typically used (mDNS, WINS,
> Hostsfile)) they also cannot use the more advanced functions of the
> system library (together with the kernel) for address selection.

Hi Bernd,

I'm not suggesting it as a "solution". I'm suggesting it as an "experiment", 
to learn more about what will happen.

> "A lot of the information used by the sorting algorithm is not easy
> to come by. It is therefore highly recommended that no program uses
> home-grown implementations of getaddrinfo(). Just use the system
> implementation."

I'm aware of this, and not recommending doing so as a real solution to the 
issue. Even though I violently disagree with the existence and usage of MDNS, 
WINS, and all of these other mechanisms, on a personal level.

> Having said that, DNS-only based implementations do have their use.
> dns,sun and dnsjava are on my list to check out :)

Excellent.

> Bernd

Matthew.



More information about the net-dev mailing list