[Fwd: Re: [gst-devel] Getting mindshare]

[Myers W. Carpenter]

I'm sure this was supposed to go here too.

-------- Original Message --------
From: Wim Taymans <wim.taymans at chello.be>
Subject: Re: [gst-devel] Getting mindshare
To: "Myers W. Carpenter" <myers at fil.org>

"Myers W. Carpenter" wrote:

Hi,

> 
>         Yesterday while trying to figure out why M$ Access didn't want to
> access it's own varibles I wander into sourceforge and was looking at
> the multimedia catagory.   To my surprise some other people have decided
> to reinvent the wheel:

I know it's getting really stupid. Not even one of them is remotely
close to
the capabilities of GStreamer AFAIK, and GStreamer is still in a pretty
rough state.

I took a quick look at the (now dead) GNOME Media Framework and it's all
too stupid. Erik really took on this project with a good deal of design,
thought and knowledge of what is involved, I think I understand why this 
is the best approach and I'm ready to defend that anytime.

My current goal is to have a real programmers manual, polish some stuff
and
start to promote things rather aggresively.

> 
>         gnoMedia player - http://sourceforge.net/projects/gno-media/
>                 gnoMedia is an attempt to bring an easy to use mediaplayer to                   the
> GNOME desktop. We have seen a lot of players around but
>                 they missed the simplicity of installing and use. Therefore
>                 this project was started.

Right, this looks good towards the end user, but what for the
programmer?

> 
>         Libmedia - http://sourceforge.net/projects/libmedia/
>                 Libmedia is the attempt to unify the APIs for the de- and
>                 encoding of media (video, audio, still images). The first goal
>                 is to create MPEG, MP3 and Ogg Vorbis decoders and provide a
>                 Gnome Widget that can play these streams.

At least 1 year behind GStreamer.

> 
>         MPEG-I video/audio library - http://sourceforge.net/projects/mpeglib/
>                 mp3, mpeg video, wav,vorbis library. supports video cds,
>                 seeking, vbr files,icecast streams, very easy to use!

oh great! MPEG-I. We can allready playback those things, no?

> 
>         Retina - http://sourceforge.net/projects/retina/
>                 Retina is a library for processing media in real-time, such as
>                 MIDI, audio, or video.
> 
>         There are others, but those are the ones that seem to be doing stuff
> almost entirely covered by gstreamer.  There is one other: libAV, but I
> started first with that project and I'm trying to have it deleted from
> sourceforge.

libAV must be merged. And that shouldn't take more than a few weeks. 

>         I'm thinking of evanglizing to some of these projects and seeing if I
> can't convice some of them to add to the effort here rather that playing
> with square things before finding round ones move better.

Good, I'm with you 100%. I want to go through some brainstorming with
you
about what and how we will do this. My approach till now was simply to
incorporate all the efforts I saw into GStreamer and extend their
usability.
But porting all those codecs by myself is not funny.

>         There are a few things I think we could also do to broadcast our
> existance a little better (not saying we should do all of them, just
> ideas):
>         - Change our freshmeat.net entry from Gnome/Core to X11/TV and Video
> (this I did already)
>         - Add Multimedia/Video and Mulitmedia/Audio to our Tove classification
> topics (it's in the sourceforge.net project admin page so Erik would
> have to change it)
>         - Change our name.  GStreamer really doesn't evoke "Media Library".  My
> sugguestion is "GNU Media Kit" or gmk for short.

I prefer something with GNOME in it. Today, anything with glib and/or
GTK+ 
automatically becomes GNOME related (well...). Can we call it GNU when
it is
tied to GTK+, GNOME, BONOBO, and all the other GNOME stuff?

>         - Polish up gstplayer and announce it on freshmeat.net.

Those are the things we need: Cool, polished multifunctional apps people
want
to use and talk about. We need to show that GStreamer is capable of much
more
than simple playback, I was thinking about pretty GUI MPEG2->DivX/MPEG
encoders etc...

Video editing would be cool too. 

>         - Come up with some short term goals, finish them and realise a new
> version.

Yes, I see these things to be done:

short term release (0.9.9 or so)
- manual
- clean up the apps (gstplay, editor)
- loud freshmeat announcements.

(1.0.0)

- bonobo component for gstplay and get the Nautilus guys use it with an
API
  like: 

    gstplay_new(<file>); gstplay_play(); gstplay_pause();
gstplay_stop();

  off course gstplay handles MPEG1/MPEG2/AVI/DivX/MP3/WAV/AU locally and
  from a network.

  blow them away with both functionality and embeding possibilities.

- GNOME control center applet. The user can choose its codecs, install,
remove
  and tune them.
- Try to get the GNOME community to accept and use it. Try to get it
into the
  main GNOME installation.

(1.0.1)

- BUG fixes
- codecs, codecs and even more codecs

(1.1.0)

- network streaming has to be perfect by then. 
- match current icecast players.

(1.2.0)

- massive parallel encoding/decoding etc.

current showstoppers might be:

- for the end user: no QoS. just one app.
- for the app developer: too much lowlevel things... need more
specialized bins.
  need better type registry.
- for the geeks: no BONOBO/CORBA...

still flaky/missing:

- xml handling
- compile/build process. (working on that)
- editor.
- properties of the codecs.
- v4l needs testing.
- just a few native avi codecs.

highlights:

- complete MPEG1/MPEG2 with subtitles.
- complete MPEG encoding: audio/video and multiplexing
- wrapper for Windows codecs (DivX, indeo,...)
- multithreading
- command line transcoders for avi2mpeg mpeg2avi mpeg2mpeg.

General:

- the current code has a lot of potential but too many things
  are in it without the full 100% testing (cothreads, XML, ...)
- need to focus on finished product instead of adding features.
- need to detach from GTK+ and use glibs 1.4 object model some day.

needs rethinking:

- type detection (too slow, need to load all the codecs)
- codec registry (same as type detection, is not scalable)

We need to focus on some of the most severe showstoppers and fix them.
We are certainly not going to solve all of the issues, we'll have to
live with them until a next release.

Tell me what you think.

Wim

-- 
The biggest problem with communication is the illusion that it has
occurred.