On 2006-08-12 at 20:41:57 [+0200], Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> wrote: > Oliver Tappe <openbeos@xxxxxxxxxxxxxxx> wrote: > > > > Well, we have BIND as part of our libnetwork.so, in the dns/ > > > > subfolder. > > > > I think Waldemar originally ported it. > > > Yes, I ported BIND, but Philippe once mentioned some better > > > resolver > > > (djbdns or tinydns?) which is less buggy and has a simpler > > > codebase: > > > http://cr.yp.to/djbdns.html > > > http://tinydns.org/ > > Yes, I personally am all for replacing bind with a simple dns-cache. > > We don't > > need the server part anyway (at least not the real thing) and the > > code has a > > bad reputation by now anyway... > > That's indeed true. Even though the client part alone is probably not > that crucial. In any way, I would not think that this is a very urgent > thing to solve, at least if BIND still runs okay in our new stack :-) Yes, there's more important stuff to do, you're right. > > I am not sure if we could use djbdns easily, as when I looked at it > > very > > briefly today, it gave me the impression of depending on other pars > > of Dan > > Bernstein's software suite. Another possibility would be to use > > dnsmasq, but > > that is GPL, I don't know if that rules it out... > > LGPL would be okay, but since all net apps would need to link against > that code, and at the net_server as well, I wouldn't think GPL is an > option. > How many other parts would djbdns need, anyway? I don't know yet, sorry. I thought I had read something in the docs along those lines, but when I tried to compile dnscache just now, it was pretty straightforward. So we might choose to switch to dnscache, later, but for now bind should do. cheers, Oliver