Some ideas for the aRts-replacement

[Marco Lohse]

Scott Wheeler wrote:

>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>On Friday 20 February 2004 11:49, Zack Rusin wrote:
>  
>
>>Just wondering what are people thinking about 
>>http://www.networkmultimedia.org . I think it's rather impressive.
>>    
>>
>
>I had been meaning to look at it for a while and pulled it down a little while 
>earlier.  I did hit at least one build problem where it was including a 
>header that it hadn't check for, but after that things built fine.
>
>Good stuff:
>
>The code is pretty nice.  It's C++ and seems pretty readable.  It's more or 
>less Qt style code.
>
>It seems that the authors are KDE users and are interested in 
>
>Current concerns:
>
>*) Out of the 4 examples in the mp3dec dir only one of them worked; simpleplay 
>specifically would really garble the mp3 for about 10 seconds and then crash.  
>This was my first impression.
>
you are right, we didn't handle that very good in the past: all stuff in 
the examples/ subdir is mainly for quickly testing new plug-ins, etc. 
Unfortunately, we then often "forget" about the one or the other 
example, and, well, things just break.

Now we have this application called 'clic' that lets you setup 
(distributed) flow graphs from a text file and most of the example are 
obsolete by now and will be removed and replaced by small text files in 
the next release.

So, if anyone of you wants to play around with NMM: please, use clic 
first (see 
http://graphics.cs.uni-sb.de/NMM/dist-0.4.1/Docs/clic/index.html )

>
>*) I'd like to hear more about the project goals -- like specifically if the 
>developers think that it's realistic for KDE's multimedia system to be on NMM 
>and if it's something they're willing to commit to supporting.
>  
>
For project goals we already set up a wiki. I am afraid, I really cannot 
tell you whether NMM is good for your needs. You have to decide that by 
yourselves. All I can do is to invite you to have a look at it. The 
other thing is that I can tell you what NMM currently already can do. 
And we can answer your questions, you just have to ask.

>*) The developers?  There are quite a few listed, but at least from the looks 
>of it they're all at the same Uni and may or may not be committed to the 
>project after graduation.  This of course also gives off the impression that 
>the goals are more academic than practical; of course I may be wrong.  It 
>also seems like (from looking at the site materials) that the system is 
>mostly maintained by Marco (again, I may be wrong -- this is just from a 
>glance at a few things).  With aRts we had a similar problem in that there 
>was really only one person that understood and maintained aRts -- after he 
>didn't have the time and/or interestet for that any longer we were in a tough 
>spot.
>  
>
You are right, currently NMM is mainly a University project. But a lot 
of people already told us that they have started to develop for NMM.

NMM is *not* mostly maintained by me. Michael is also working at the 
University within the NMM project. Among many other things, he is 
currently in charge for all the releases. And most important: There are 
a lot of motivated students working in the project. You might be right, 
students mainly work within the NMM project to do a Master or Bachelor 
thesis. How much they want to contribute to other things really depends 
on what the student wants. I think this worked really good so far: we 
have always some students leaving the project, other students start to 
develop things for NMM. That is pretty much the same in other "pure" 
Open Source projects.

I cannot tell you how much each of us will contribute to "the" KDE 
multimedia architecture. I think we will need some people working only 
on NMM, some working on KDE, some in between, then. I guess this will be 
the same whatever "external" project you will finally choose (e.g. 
gstreamer, MAS, mplayer, ...).

>*) At least my first impression is that it's not as mature as other options.   
>This is somewhat reinforced by the lack of "real" applications that seem to 
>be using it.  I may try porting JuK's relatively simple Player interface to 
>it just to get a feel for the API and stability.
>  
>
That is maybe one of the myths of NMM: "the lack of "real" applications"

Just take a look at the Multimedia-Box: 
http://graphics.cs.uni-sb.de/NMM/Status/MMBox/index.html
it is TV, a DVD player, a CD player, a grabber, a distributed 
transcoder, a general player with play-list,..., all within one big 
application that allows multi-tasking.

If you are too lazy to test it yourself: watch this video 
http://graphics.cs.uni-sb.de/NMM/Status/MMBox/index.html#videos or come 
an visit us at CeBIT.

If this is not "real" what else?

I will totally agree that there are systems on which NMM/the 
Multimedia-Box does not work properly. I will also agree that there are 
still bugs in the code. The main reason for this is: we do not have a 
big user base, things were mainly tested on our machines. And often we 
simply cannot reproduce bugs because they only occur when using some 
piece of hardware, some DVD, etc. that we do not have.

But: this does not mean that things work in general and can be fixed.

The other big myth about NMM: "it is network transparent, therefore 
there is a lot of overhead and code I do not need for my locally running 
application"

In NMM, networking can also be plugged in like any other plug-in, but is 
not part of the core architecture. You will have no overhead for all 
locally running parts of the application.

>*) The development list is pretty inactive and it seems like most of the stuff 
>there is really build support and whatnot -- this may be because the 
>decisions are made in the lab where everyone's around or mostly by one 
>person...  Also I didn't notice an IRC channel, cvs list, public bugtracking 
>system, etc...
>  
>
well, we do have a cvs list for sourceforge, we do have a forum, we do 
have wiki pages, things just get started...

All the other things like IRC, etc. : well, I think, if people will need 
these things, someone will just simply start it. Currently I do not see 
a reason to use IRC to talk to the guy who sits next to me all day ;) 
but I could use it if someone wants to talk to me that way.

>I'd like to hear more about it -- notably Marco's thoughts on these issues.
>  
>
Ok, my summary: NMM is a project developed and maintained by many 
people, mainly from Saarland University. We would like to discuss things 
with you. We currently cannot make any claim on how much everyone of us 
will support "the" KDE multimedia architecture. But: NMM is out there, 
it is a stable multimedia architecture, it has some basic features that 
other projects that are supposed to be "more stable" do not have (e.g. 
DVD support). It is academic, but it is also useable and practical.

So long, have fun, Marco.