Comments on article "Coherence and Nautilus, UPnP a bit more on the desktop"

lightyear on Coherence and Nautilus, UPnP a bit more on the desktop

lightyear — 2008-10-29T13:14:17.000002+02:00

I have to agree with dev here.

As far as I understand Coherence it is not a simple UPnP-Stack allowing an application to access the UPnP world. In contrary I see its place an always in background running service on my system. And when I fire up my favourite application (may it be Totem, Elisa or Whatever) it can make use of that service. Currently this can be done over DBus but there are also other concepts out there that just need to be implemented at some point.

But that is only one service that Coherence provides. Let me explain what else I see in here. Lets think about my nice new TV. It speaks this weird protocol (or stack if you want) called UPnP. Over that I can access the media in my network - Great. But you know what? Most of my pictures are hosted on flickr and on an external harddrive I don't want to plug every time. So what do I do? I start Coherence. Because Coherence does not only provider services to the applications running on that desktop, it also allows libraries and WebAPIs to make their content available over UPnP. And I don't have to write a specialized application for that every time. Coherence has a flickr backend (and I think others will follow). So I can watch my flickr photos on my TV using UPnP even though the TV does not have any particular support for that specialized service. (Since the recent changes that are shown in the Video this Blog entry is about I can even upload pictures to flickr using Coherence or UPnP only. - AWESOME!)

So Coherence is not so much about providing a UPnP stack to some application. It is about providing access to the media (of/to application and of/to UPnP devices).

Just my 2 cents. ben

dev on Coherence and Nautilus, UPnP a bit more on the desktop

dev — 2008-10-28T23:46:25.000001+02:00

Eric, Coherence is _not_ only another UPnP stack, so it is not that easy to be compared with GUPnP.

Let me elaborate this a bit.

Imho UPnP isn't there for itself. The reason for UPnP is to provide a simple way for applications to share services and data. So it is actually nothing more than some additional RPC concept. And a not even very well designed one and maybe even more difficult to handle.

So there is a very small amount of people out there waking up in the morning with the strong intention to write today for example an UPnP MediaServer.

On the other side there are a lot of people who have digital media - let's concentrate on that, the UPnP A/V subset, for the moment - and quite a few out of them write great applications that deal with that media. To display it, organize it, transfer or share it,...

Now if it would be only about your laptop and your desktop, nobody would spend a thought about UPnP.

But a few years ago some usually egocentric companies in the Consumer Electronics world made a strange decision. They agreed on using UPnP A/V to let their so far closed devices talk to each other.

And suddenly, roughly 10 years later, we have there such things like the PS3 or DLNA/UPnP enabled TVs.

And now we come to probably the only reason why we should spend a second thought on UPnP - if we want to work with these devices, which will show up in our households more and more, if we want to let them access our digital media, we need to provide UPnP functionality to these applications that handle our media, and to the people that work on them.

One option is to use a dedicated UPnP MediaServer for this - but isn't that ridiculous? There are awesome applications already there, which we use on a daily basis to deal with our media, why should we start using another one just for getting that to your TV?

Of course we could throw several UPnP stacks at the application developers and demand one of them to be integrated in their app, and if we are lucky and the developers are masochistic enough they might consider doing that.

Selecting the stack that fits best to their app, trying to understand how UPnP works, dealing with the errors and the peculiarities of that stack. Just to find out in the end that it maybe works with the Sony PS3, but not with a Sony TV.

Now Coherence has a completely different approach.

Let's say there is an applications that deals with audio media files. Now in the UPnP-stack world, it would have to do all necessary things to speak UPnP. With one stack more, less with some other.

With Coherence that application just(*) needs to say: I want to be a MediaServer, and this is the data I can provide.

Or if we look at a media player: I want to be a MediaRenderer, and I can play, stop, pause, skip.

Coherence does provide docks for these applications -> http://coherence.beebits.net/wiki/ArchitecturalOverview - so one Coherence instance, or daemon, can handle several MediaServer or MediaRenderer apps, or ControlPoint clients, or...

My initial idea was to _not_ have every application to integrate UPnP functionality in itself - replicating every time the work done already in some over app - nor was it to replicate the functionality of these applications on the Coherence side.

But to provide an easy to use interface for everyone - be it Python, Mono, C, or whatever shows up next year - and allow the apps to concentrate on their primary mission.

And I still believe that this is the way to go - not overestimating the importance of 'a stack' - but providing an easy way to enable the applications we actually use.

And finally one word to Zeeshan - pls provide a fair comment about using D-Bus for providing that interface resulting in a performance issue in a real-world situation. Or stop propagating your own lack of judgement as a statement of fact.

(*) of course it is not yet that simple, but if you look at the work we've done so far with Ampache, Elisa, DVB-Daemon, Rhythmbox, Totem,... you'll see that we are pretty close, and getting nearer every time.

phil on Coherence and Nautilus, UPnP a bit more on the desktop

phil — 2008-10-28T15:02:08.000001+02:00

Sorry, misread indeed... We'll see when GUPnP has Python bindings then. But if I am a Python application developer looking for a portable UPnP library, I'll choose Coherence. Pure python will work on any OS where Python is available, minus some quirks and bits to sort out ;)

Anyway, people have freedom of choice and we all agree on that!

Zeeshan Ali on Coherence and Nautilus, UPnP a bit more on the desktop

Zeeshan Ali — 2008-10-28T13:25:36.000001+02:00

Actually If you read what i've written above carefully, I am talking about giving application developers the freedom to choose their favorite programming language, which includes python. That is the reason I will be the happiest person on the planet when we have nice python bindings for GUPnP. I am really sorry if you don't see my objection on mis-use of python but if you still don't get it, maybe it's not me who is being dogmatic/religious here.

phil on Coherence and Nautilus, UPnP a bit more on the desktop

phil — 2008-10-28T10:16:58.000001+02:00

Zeeshan,

I don't want to be aggressive either, but I'm getting tired of you always whining about Python ;) You don't like it, that's fine, we know now. So what about being a bit more tolerant and let people use the programming langage they want? :)

Zeeshan Ali on Coherence and Nautilus, UPnP a bit more on the desktop

Zeeshan Ali — 2008-10-28T10:05:39.000001+02:00

Oops Sorry to post the same comment thrice but my browser kept on showing me preview even after hitting "postComment" twice.

Phil! please delete the redundant comments.

Zeeshan Ali on Coherence and Nautilus, UPnP a bit more on the desktop

Zeeshan Ali — 2008-10-28T10:02:50.000001+02:00

Eric! You seem to have got everything correct, except that you don't need to worry about who wins in the end. Being a main developer of GUPnP, I must admit that Frank's work has been a source of motivation (and in some cases inspiration). I don't know of any "war" between pigment and clutter either. :) IMO, once GUPnP offers all the features that Coherence does and provides nice bindings to most important high-level languages, I am sure it will be obvious which will become the (defacto) standard UPnP stack under GNOME (We've already won the battle on Maemo).

Phil! Sorry to sound aggressive but you hardly said any good thing about GUPnP so couldn't resist. :) The only way non-python apps can currently use Coherence is through D-Bus and in many cases doing IPC for RPC is a terrible idea wrt performance (e.g farsight firing holes though firewalls). There is a good reason why most libraries/frameworks in GNOME are still written in C, despite the fact that the developers are much more comfortable with Python.

phil on Coherence and Nautilus, UPnP a bit more on the desktop

phil — 2008-10-27T23:57:31+02:00

I think it's always a good thing to have some diversity. Frank puts a lot of efforts on Coherence and has the vision to build bridges between various kinds of applications and spread the usage of UPnP... Now, if Coherence isn't part of GNOME, is that really a problem? There's not only GNOME in this world. I see more and more applications using Coherence, being part of GNOME, or not. I think the point is to spread the use of UPnP and that's what GUPnP and Coherence are doing, which is a good thing.

Now about Pigment / Clutter, yeah the war (was there one, really?) has ended a while ago. I think one of the reasons is that there's mostly only one engineer working full-time on Pigment, whereas there's a whole team behind Clutter and people hacking on various libs linked to Clutter, this is a huge job. And yeah if Clutter ends up being integrated in GNOME that's good, the important thing is to make the desktop evolve and that's a key to the evolution ;)

Eric Miller on Coherence and Nautilus, UPnP a bit more on the desktop

Eric Miller — 2008-10-27T19:51:19+02:00

Being a close follower of planet GNOME, I had been seeing two completely different UPnP stacks cooking up: Coherence and GUPnP. While Coherence seems to be way ahead of GUPnP when counting featuresets and stability, I can't help but wonder how will it possibly compete with GUPnP in the long run since GUPnP is developed by a (now) big company and is supported by another big company: Nokia.

Keeping that in mind and other (minor imho) facts like that it is written in C and specifically designed for GNOME world, it seems quite like the competition between Clutter and pigment and we have all seen who won that war, despite the desperate efforts of pigment team to convert most code to C in the end, which doesn't seem to be happening in case of Coherence.

I am sorry to be so negative but I am saying this in hopes that I am missing some important facts here and you'll point me to them. :)