Re: [Rhythmbox-devel] ANN: RhythmboxLIRC -- a program to control Rhythmbox with a remote control

[Jonathan Matthew <jonathan kaolin hn org>]

On Fri, Sep 23, 2005 at 04:13:47PM +0200, Christophe Fergeau wrote:
> Le samedi 24 septembre 2005 à 00:11 +1000, Jonathan Matthew a écrit :
> > LIRC support is self
> > contained, so this might be helpful as a use case for a plugin system or
> > whatever.
> 
> I'll keep repeating myself, but lirc support (and last.fm support for
> that matter) are the kind of things that are perfectly doable using the
> bonobo/dbus interfaces ;)

While I agree, I think there's a reasonable case for including lirc
support in rhythmbox itself.  It's a small amount of fairly brainless
code that does nothing if the user doesn't run lirc.  If enabled, it
adds one dependency on a small library that totem already depends on (at
least in debian and ubuntu).  The lirc client library interface is
stable.

As I see it, the tradeoff in pushing functionality like this out to
bonobo/dbus clients is:

- Clients can be written in higher level languages (most are in python)
- Clients don't add dependencies to rhythmbox
- Less rarely-tested and probably-unmaintained code in rhythmbox

vs

- Users have to find the clients for themselves.  Some are packaged by
  distributions, which helps.  Otherwise, listing them on the website
  is good, but links do rot sometimes.
- Users have to run client programs manually in most cases.  This can be
  fixed, but it's not easy to do this in a user-friendly manner as far
  as I can tell.  For me at least, this results in just not using the
  client, rather than making the effort to run it.

I think there are some cases, such as lirc support, where the tradeoff
isn't worth it.  I can't see anything on
http://gnome.org/projects/rhythmbox/utils.html that I think could
reasonably be built in, though.

In any case, if people are going to continue to make patches like this
(and it seems they are), then I think they should at least be good
patches.

-jonathan