Re: Working with a local DNS cache

[Adam Langley <agl imperialviolet org>]

On Fri, Aug 21, 2009 at 8:04 AM, Dan Williams<dcbw redhat com> wrote:
> Well, Marcel had some valid issues with dnsmasq, and others may want to
> use bind, so in the end we probably want a GInterface for this and then
> have classes that implement the GInterface for each of the specific
> caching name daemons.  I don't really have a problem with that; we'll
> probably have to add arguments and a config file option for which one
> you want to use, because people may want to use dnsmasq underneath
> chromium too and have chromium still use 127.0.0.1.

I only wrote the code using org.chromium.LocalDNSCache because that's
what I have running here. For the final version, I think it should use
the dnsmasq interface[1]. I've no desire to fork interfaces with out a
good reason. (The dnsmasq interface isn't very pretty, but I don't
know if that's a good enough reason.)

I'd be happy to write a GInterface to abstract the configuration if
that's what's needed. But shouldn't DBus *be* that interface? I
suppose if BIND isn't going to implement a DBus interface then we have
little choice, but rewriting BIND configs is a rather scary
proposition. Would people want a templating system in order to have
NetworkManager write a config which preserved their BIND preferences?

[1] http://www.thekelleys.org.uk/dnsmasq/docs/DBus-interface

> No problem; everyone has their own.  A few of the guidelines are
> documented in the CONTRIBUTING but I see now that I should extend that
> significantly.
>
> A few style comments...

Thanks! Hopefully I've fixed up those style issues. (Note: rebased on
top of up-to-date tree). Also updated the CONTRIBUTING file with some
of your style points:

http://github.com/agl/NetworkManager/commits/agl


Cheers

AGL

-- 
Adam Langley agl imperialviolet org http://www.imperialviolet.org