Hi, Laurent Bercot wrote on 2011-11-24: > > This patch integrates the nss-mdns approach into uClibc's resolver, > > so getaddrinfo() and getnameinfo() calls get redirected to a running > > avahi-daemon before trying to resolve via unicast DNS. > > Rule of Separation: Separate policy from mechanism; separate interfaces > from engines. (http://www.faqs.org/docs/artu/ch01s06.html) > > If uClibc is going to support mDNS, at least please support it in a > uClibcesque way. Do not copy the glibc mechanism, and do not assume users > are going to run avahi-daemon, because they might very well be running > other mDNS software with a totally different mechanism.
the resolution is done with ASCII communication via a unix domain socket. I consider this to be adequately abstract and simple to be accepted as an abstract interface that could be supported by other mDNS software. > Some of us are using uClibc *specifically* to get away from retarded > glibc mechanisms. :) The commit text is misleading. The glibc way is to do the resolution via libnss. This patch uses abstract domain socket communication, which can easily be integrated in any mDNS software. Avahi as the de-facto standard for mdns in the Unix environment has a reasonably abstract interface and the nss-mdns makes use of this interface to integrate mDNS in glibc. That's why I mentioned it. Sorry for the nebulosity, Niko _______________________________________________ uClibc mailing list [email protected] http://lists.busybox.net/mailman/listinfo/uclibc
