[MPlayer-dev-eng] MMS support in mplayer begun

[Bertrand Baudet]

On Thursday 06 December 2001 01:21 am, you wrote:
> Hi, all. I have started on MMS support in mplayer.  I have it at least
> connecting to TCP MMS streams.  I have included the patch. For some
> reason, if I try using "cvs rdiff 120601 main", I keep getting a whole
> messful of parsed_root_directory errors. So,  I  was only able to run
> "cvs diff -NR" to get a file. I had to edit the file to get a proper
> patch file. But, it didn't include the new file I put in.

Have you RTFM: mplayer/DOCS/tech/patches.txt ?

> So, I have included both the patch and the new file.  I do have my test
> mms stream being connected.  I do send it the correct header. I have yet
> to get it to start the actual data packets. But, it should not affect
> anything else in mplayer. Let me know what you think.

I'm not sure if I will be able to have a look a this before the week-end.
I had a quick glance at the patch and I have several comment at the first
look:
- First of all, are you award that even if you do the mms(t/u) implementation,
  that will not work better than the current HTTP streaming implementation?
  The network streaming part is working fine AFAIK. To make the ASF
  streaming work properly either with HTTP(from WMS) or with MMS, the
  ASF core code and maybe other source code in libmpdemux need to be
  changed/fixed. I'm still investigation that. I have some ideas on what's
  going on. I wanna check this first try to make a dirty hack to see if it
  can make the streaming work better.
- Have you read this: http://freespace.virgin.net/sdpman.sdp/ ?
  Especially the document that the guy wrote?
  http://freespace.virgin.net/sdpman.sdp/MMSprotocol.doc
  I had a glance at it yesterday night, and it look interesting, it give
  some background on ASF streaming, and it try to explains how MMS(T/U)
  works.
- I don't really like the MMSC_* definition in mmst.h, we usually do that 
 when we first reversed the protocol and we know nothing about it. But once 
 we start knowing more about the fields involved, we have to make the proper 
 structures. It makes the code easier to read and debug. And it will be much  
 more easier after that to make it endian independant. 
 Please have a look at the SDP doc above and try to see if you can make some 
 structures out of this. Hopefully after that you can remove a lot of the
 unused helper functions.

That it for now, I didn't try to apply to patch yet, so I will give you some
other feed back after I did it.
Like I said, I may not be able to test the patch before the week end, so if
you change it and if you make some clean up, please send the new patch so
I can have a look at this this week end.

Bertrand