Some ideas for the aRts-replacement

[Marco Lohse]

Hi there,
I am one of the developers of NMM and I would like to make some comments.

Allan Sandfeld Jensen wrote:

>On Friday 20 February 2004 12:54, Gioele Barabucci wrote:
>  
>
>>Zack Rusin <zack <at> kde.org> writes:
>>    
>>
>>>Just wondering what are people thinking about
>>>http://www.networkmultimedia.org . I think it's rather impressive.
>>>      
>>>
>>It is very impressive:
>>from http://www.networkmultimedia.org/NMM/Docs/index.html
>>
>>Things we really need:
>>- Decoding/Encoding/Output plugins (many already implemented)
>>- Synchronisation
>>- Threads support
>>
>>Things is nice to have:
>>- LGPL'd + some GPL plugins
>>    
>>
I think nearly all of our plug-ins and all application are GPL.

>>- Distributed/net media
>>- Serialization
>>- Doxygen generated documentation of API and a clean C++ coding style.
>>
>>Three are my fears:
>>- Is it light enought? I mean, will it take ~1% of CPU time while decoding
>>an MP3 or 30%?
>>
I recently ran a comparison for video playback for mplayer and NMM's 
clic, the results:
http://graphics.cs.uni-sb.de/pipermail/nmm-dev/2004-January/000680.html

>>- How simple is to create/integrate plugins? From the doc is seems rater
>>simple, but you can't say anything untill you try.
>>
we used NMM in two courses at Saarland University: students got 1,5 
hours of introduction and were able to develop plug-ins for 
VISCA-cameras, icecast, realserver, divx,  imagemagick, ... here are the 
results:
http://graphics.cs.uni-sb.de/Courses/ws0102/Multimedia/#Projekt
http://graphics.cs.uni-sb.de/Courses/ss03/Multimedia/projekt/MMProject.html

>>- How much will they support an external big and annoy user like KDE? :)
>>    
>>
>
>If we choose to adopt it for KDE4, I guess we would have to support it 
>ourselves. I tried to download and build it a few months ago and I didnt 
>succed, after fixing around 10 build-bugs I still hadnt even made it even 
>configure (just means they have a very narrow userbase at the moment). On the 
>good side it seems to be somewhat tied to KDE-technology already, and unlike 
>GStreamer it has builtin network-transparency.
>  
>
If you found 10 build-bugs, it would be great if you could send us a 
patch. On the other hand, people have reported that they successfully 
compiled NMM on many different platforms, mostly Linux, but also other OS.

NMM has a lot of external dependencies. And NMM is already a lot of 
code. That is why people have problems building it. We try to do our 
best to improve this situation: we have a list of needed external 
libraries, pre-compiled external libraries, binary packages for NMM, 
nightly builds of our sourceforge CVS, ..., see 
http://graphics.cs.uni-sb.de/NMM/Download/index.html

One important thing: the base system is very small: e.g. we run NMM on 
Linux iPAQs, e.g. for watching TV that gets transcoded on a PC. The 
"size" of NMM mainly comes from the plug-ins.

A personal remark: NMM is a research system. This does not mean that it 
is not stable. This does not mean that we do not want KDE people (or 
someone else) to use it. It only means that we (or, at least I 
personally) still see it as a research project. In general, a research 
project could be integrated in "big things" like KDE.

One last remark: we exhibit at CeBIT 2004. So if anyone is interested in 
seeing NMM in action, you are heavily invited to join us: future parc, 
hall 11, booth E 30 (Saarland Forschungsstand). We will post further 
details on our website and mailing-lists soon.

So long, have fun, Marco.