Re: Moving to *Avahi* over howl

[JP Rosevear <jpr novell com>]

On Tue, 2005-09-20 at 11:48 +0200, Alexander Larsson wrote:
> On Mon, 2005-09-19 at 12:44 -0400, JP Rosevear wrote:
> 
> > > Since DNS-SD is pretty close to the system-level its not at all
> > > unreasonable that we should use the system-specific implementation. On
> > > the other hand, Avahi clearly looks like the best choice on Linux and
> > > other free UNIXes, so almost all Gnome users would use that, and would
> > > pay for the extra layer unnecessarily.
> > 
> > You also arguing against using something "uncondtionally" above, which
> > is exactly what you are promoting here.
> > 
> > It would seem in cases where we've had different implementations in the
> > past, or multiple implementations of the same thing, this would suggest
> > an opportunity for an abstraction layer.
> 
> I'm not arguing that an abstraction layer for this is inherently bad. In
> the end it might be the right thing to do. But there are several open
> issues here, such as:
> 
> In what library would the abstraction layer live?
> 
> Adding a new library for something like this does cause a form of bloat,
> both in build complexity and the need to load/link more libraries.
> Adding it to an existing library is complex since that would add more
> dependencies to that library which isn't always very nice.
> 
> Who would design and write it? Who would maintain it?

I can't answer these questions for sure, although there have been some
people proposing on the list.

> Do all implementations have the same features, or do we have to cater to
> a least common denominator?
> 
> It seems like this is not really the case, at least some comments in
> this thread says Avahi is more powerful. A least common denominator
> gives us a less powerful API and possibly a less featurefull desktop.
> And there is a risk that apps start using Avahi anyway as they need some
> feature, and then we're not portable anymore.

This is exactly whats occurring with the cross platform status icon api
though isn't it? (I don't think I'm for the lowest common denominator
there either, but I haven't read all the arguments).

-JP
-- 
JP Rosevear <jpr novell com>
Novell, Inc.