I'm just wondering, what is the goal of c-ares? Is it simply to be a libcurl
resolver? Or is the goal a general purpose resolver that takes over where
ares left off? I ask because there are many features of DNS that probably
aren't useful to curl/libcurl, but probably would be useful to others. I
began wondering about this with the post about SRV records. If we add SRV
record support (which I really think we should), should we then also add
support for other records? Like I could see MX support being useful for some
people as well? Of course, the ares_query function already makes this
possible, but, I think, if I'm making an application that uses name
resolving, I shouldn't have to know the DNS protocol, the library should
hide all of that from me.

Basically, what I mean is, with some of the features suggested (IPv6,
nsswitch.conf, SRV support) c-ares has the potential to grow into a very
versatile library. Is that the goal? Or is it simply to have a minimalistic
library to provide libcurl with what it needs?

Dominick Meglio
Received on 2004-03-19