Zeroconf in KDE4

[Jakub Stachowski]

Dnia piątek 23 luty 2007, Hubert Figuiere napisał:
> Aaron J. Seigo wrote:
> > On February 22, 2007, Hubert Figuiere wrote:
> >>> If it makes any difference, Apple's mDNSResponder is now licensed under
> >>> Apache 2.0/BSD (the shared library part is BSD).
> >>
> >> Why just not using Avahi?
> >
> > perhaps you could explain the benefits of Avahi compared to mDNSResponder
> > for those of us who don't follow either project?
>
> 0/ it is a Free software community project, unlike Apple's mDNSResponder
> that is mostly Apple project released as open source. That means Avahi
> developement is likely to be much more responsive to the requirements
> leveraged by Free Desktop implementation that Apple's.
>
> 1/ it is LGPL licensed. I know that mDNSResponder licensing has been
> resolved...
>
> 2/ it has a D-Bus interface. KDE4 already switched to D-Bus.
>
> 3/ it has compatibility libraries. That means existing code is supposed
> to work with Avahi without trouble. Seemless migration, and you still
> have a plan B to support *non-free* platform, like mentionned earlier.
>
> 4/ it is used by Gnome (or at least several of their mainstream
> applications), and that later point to strictly emphasize the common
> platform to implement Free Desktop. That also means that you don't have
> to incompatible implementations of the same desktop-independent stack if
> you run both Free Desktop on the same machine. Yes there are a lot
> people that end up runing both because of their choice of applications.
> Combination that comes to mind are amaroK under Gnome for example.
>
> More info from the project itself:
> http://avahi.org/wiki/AboutAvahi
>
> All in all, I think it is a better choice to use Avahi.

Another points to consider:

1/ Avahi does no support wide-area service publishing. But on the other hand: 
is there anyone who actually uses it?

3/ This is from bonjour-dev mailing list (question was about status of 
wide-area support and NAT traversal on POSIX platform, the one who responds 
is Marc Krochmal, lead developer of this stuff):

>> 7. And finally, as I can see, the mac implementation is much better or
>> advanced. It is clear to me that there are many platform  
>> differences, but
>> In an environment where many services are hosted on posix machines,  
>> I am
>> dependant on a complete implementation. I would be glad if the team  
>> could
>> spend some time on the posix code to bring it on the same level  
>> like the
>> mac code is.

>Yes, unfortunately we lack the resources to keep the POSIX  
>implementation working at the same level as the Mac and Windows  
>versions, and rely more on the developer community to help us out.   
>We're constantly accepting patches to improve POSIX, and we have a  
>butch a patches waiting to be integrated that we hope to do soon in  
>order to release a new stable build.  Stay tuned.

As you can see Avahi will move forward faster than mdnsd on Linux and BSD 
platforms. Also Avahi being much more popular there helps. On the other hand 
Windows and MacOSX users will be better served by mdnsd.

3/ If you find a bug in Avahi you can count on quick response from developers. 
If you send them a patch it will be checked and commited in day or two. 
Compare it with Apple's way: you send a patch to the mailing list then wait, 
wait and wait some more. This may be because I sent patches fixing problems 
in linux port which is hardly a priority for Apple.

>
> Just my $.02
>
> Hub