Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: MythTV: Dev

Re: [mythtv-commits] mythtv siparser changes....

 

 

MythTV dev RSS feed   Index | Next | Previous | View Threaded


stuarta at squashedfrog

May 3, 2005, 4:24 AM

Post #1 of 11 (1649 views)
Permalink
Re: [mythtv-commits] mythtv siparser changes....

On Mon, May 02, 2005 at 05:45:02PM +0000, mythtv[at]cvs.mythtv.org wrote:
> ----------------------------------------------------------------------------
> Changes committed by danielk on Mon May 2 17:41:05 2005
>
> Modified Files:
> in mythtv/libs/libavcodec:
> parser.c
> in mythtv/libs/libavformat:
> utils.c
> Log Message:
>
[snip]
> nor the changes the SIP parser. The SIP parser changes need to be reviewed by
> someone familiar with DVB. It seems to be a hack to allow video and audio to
> be decoded from the same type id (0x06), if there is one video stream and it
> is listed before the audiostreams.
>

That's the freeview /xtraview patch that was sent to the list a couple
of weeks ago. There is however a better way to do it than the original
patch ;-)

Therefore, Daniel, it's not relevant to the patch in question....


Stuart


rtjacob at earthlink

May 3, 2005, 7:47 AM

Post #2 of 11 (1627 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

Quoting Stuart Auchterlonie <stuarta[at]squashedfrog.net>:
> That's the freeview /xtraview patch that was sent to the list a couple
> of weeks ago. There is however a better way to do it than the original
> patch ;-)

This patch is on the frontend not the backend, so it wouldn't apply to this set
of code anyway..

Taylor
_______________________________________________
mythtv-dev mailing list
mythtv-dev[at]mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev


mythtv-dev.spam at isismanor

May 3, 2005, 9:45 AM

Post #3 of 11 (1628 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

Stuart Auchterlonie wrote:
<snip>
> That's the freeview /xtraview patch that was sent to the list a couple
> of weeks ago. There is however a better way to do it than the original
> patch ;-)

What's the better way?
_______________________________________________
mythtv-dev mailing list
mythtv-dev[at]mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev


stuarta at squashedfrog

May 3, 2005, 1:32 PM

Post #4 of 11 (1618 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

On Tue, May 03, 2005 at 05:45:12PM +0100, Neale Swinnerton wrote:
>
>
> Stuart Auchterlonie wrote:
> <snip>
> > That's the freeview /xtraview patch that was sent to the list a couple
> > of weeks ago. There is however a better way to do it than the original
> > patch ;-)
>
> What's the better way?

Okay, the old version was messy because it assumed that the
private data streams were either video or audio. I identified an
id byte in the private streams which relate to video and audio.

The same byte can also be used to identify subtitle descriptors.
Someone else is already looking into dvb subtitles.


Oh and there's a patch as well.....
Stuart
Attachments: my-freeview.diff (0.95 KB)


stuarta at squashedfrog

May 4, 2005, 9:26 AM

Post #5 of 11 (1618 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

On Tue, May 03, 2005 at 09:32:06PM +0100, Stuart Auchterlonie wrote:
> On Tue, May 03, 2005 at 05:45:12PM +0100, Neale Swinnerton wrote:
> >
> >
> > Stuart Auchterlonie wrote:
> > <snip>
> > > That's the freeview /xtraview patch that was sent to the list a couple
> > > of weeks ago. There is however a better way to do it than the original
> > > patch ;-)
> >
> > What's the better way?
>
> Okay, the old version was messy because it assumed that the
> private data streams were either video or audio. I identified an
> id byte in the private streams which relate to video and audio.
>
> The same byte can also be used to identify subtitle descriptors.
> Someone else is already looking into dvb subtitles.
>

I forgot to mention that there is a legitimate use for this patch.
BBC Parliment channel carries 1 video stream and 3 audio streams.
The video stream contains three smaller streams comprising of
2x BBC News 24 Interactive and 1x BBC Parliment.
The 3 audio streams correspond to each video stream.

At the moment you have to use +/- to change the audio channel.

At some point in the future these will work like they do on a
set top box, ie. The interactive stuff will all work, but that's
a work in progress. (I'm working on it, but it'll take a while..)


Stuart


rtjacob at earthlink

May 4, 2005, 11:39 AM

Post #6 of 11 (1610 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

Quoting Stuart Auchterlonie <stuarta[at]squashedfrog.net>:
> I forgot to mention that there is a legitimate use for this patch.
> BBC Parliment channel carries 1 video stream and 3 audio streams.
> The video stream contains three smaller streams comprising of
> 2x BBC News 24 Interactive and 1x BBC Parliment.
> The 3 audio streams correspond to each video stream.

How does the xtraview patch fix this? Are the audio streams mistyped as
something else (Sorry it's been a while since I looked at it)..

> At the moment you have to use +/- to change the audio channel.

And how would you change this? Play all 3 at once? :)

> At some point in the future these will work like they do on a
> set top box, ie. The interactive stuff will all work, but that's
> a work in progress. (I'm working on it, but it'll take a while..)

For True interactive services would you not need access to all PIDs in the
transport stream? What about rendering the interactive pages?

Taylor
_______________________________________________
mythtv-dev mailing list
mythtv-dev[at]mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev


stuarta at squashedfrog

May 4, 2005, 1:26 PM

Post #7 of 11 (1629 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

On Wed, May 04, 2005 at 02:39:02PM -0400, Taylor Jacob wrote:
> Quoting Stuart Auchterlonie <stuarta[at]squashedfrog.net>:
> > I forgot to mention that there is a legitimate use for this patch.
> > BBC Parliment channel carries 1 video stream and 3 audio streams.
> > The video stream contains three smaller streams comprising of
> > 2x BBC News 24 Interactive and 1x BBC Parliment.
> > The 3 audio streams correspond to each video stream.
>
> How does the xtraview patch fix this? Are the audio streams mistyped as
> something else (Sorry it's been a while since I looked at it)..

Short Version: Yes.

Long Version:
Effectively they deliberately mistype the video and audio streams
and then use the mheg application to select the streams that are of
interest. The patch "fixes" this because it is classifying streams
as either audio or video (depending on the id byte it finds), rather
than having an application say "I want PID x and y." I strongly
suspect the application would do exactly what the patch does.

BBC Parliment is actually an mheg application that loads the
video out of a private stream, overlays the drawings from the
application (some text, some coloured boxes etc) and selects
the appropriate audio channel from another private stream.

BBC News interactive is similar. You already start off in the
interactive application, which then pulls in the video stream
then depending on which "video" you want it selects the
appropriate audio channel for that "video"

I say "video" because they have a headlines rolling loop and
another loop which the user can select between, when there is
actually only one video stream (as explained before...)

It seems to be a fairly common way of having a channel startup
with an interactive application.

>
> > At the moment you have to use +/- to change the audio channel.
>
> And how would you change this? Play all 3 at once? :)

By finishing the rest of the interactive / mheg stuff. :-)

>
> > At some point in the future these will work like they do on a
> > set top box, ie. The interactive stuff will all work, but that's
> > a work in progress. (I'm working on it, but it'll take a while..)
>
> For True interactive services would you not need access to all PIDs in the
> transport stream? What about rendering the interactive pages?
>

It gets quite messy. Hmmm, where do I start....

Okay, you need access to SOME of the PIDs in the transport stream.
There are a number of different stream types that we are interested
in for interactive tv. The primary one is 0x0B, which is assigned
to DSM-CC. There are commonly several PIDs on one TS which are of
this type...

DSM-CC translates as the Object Carousel. All MHEG applications are
delivered through DSM-CC as well as get all of their data through it.
The object carousel collects and caches all the bits and pieces
that are transmitted, which includes all MHEG related stuff,
firmware upgrades for STBs etc... The UK MHEG profile defines what
is mandatory, optional etc....

Anyway, the Object Carousel would need to reside in the backend
and be available to the frontend (This is a **MAJOR** TODO....) while
the MHEG rendering engine would need to be on the frontend.

While we are talking about major TODOs there will be a need for a
mechanism for the frontend to tell the backend which PIDs it wants.
This is because the interactive tv commonly needs to 'select' what
the viewer is interested in looking at.


Aren't you glad you opened that tin of worms.... :-/


Stuart


brejc8 at vu

May 5, 2005, 12:14 AM

Post #8 of 11 (1632 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

Yeah the original patch is patchy but I don't think this patch is perfect
either. Basically this picks off the first descriptor_tag. I did run a few scans
to look for such a 'tell' but it isn't that easy. I hacked about the dvbscan
util to tell me more info about the streams it doesn't know about (I can send
you a patch). From what I have seen yeah the audio streams have an 0x0A tag on
them and they are easy to recognize. But the both the Audio and Video streams
have the 0x52 tag on them (the audio usually has the 0x0A first though). So the
patch seems correct unless they change the order of these tags.

Stuart Auchterlonie wrote:
> While we are talking about major TODOs there will be a need for a
> mechanism for the frontend to tell the backend which PIDs it wants.

I think this would be much simpler done if the backend simply records all
streams given in the channel description. The front end can then pick and choose
which streams to play. This will allows you to record interactive content,
teletext, subtitles... and then maybe later when viewing select the stream you
want to listen/watch though the interactive interface or a manually cycle
through all (suspected) audio/video streams.
--
Charlie Brej
APT Group, Dept. Computer Science, University of Manchester
Web: http://www.cs.man.ac.uk/~brejc8/ Tel: +44 161 275 6844
Mail: IT302, Manchester University, Manchester, M13 9PL, UK
_______________________________________________
mythtv-dev mailing list
mythtv-dev[at]mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev


stuarta at squashedfrog

May 5, 2005, 2:56 AM

Post #9 of 11 (1621 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

On Thu, May 05, 2005 at 08:14:31AM +0100, Charlie Brej wrote:
> Yeah the original patch is patchy but I don't think this patch is perfect
> either. Basically this picks off the first descriptor_tag. I did run a few
> scans to look for such a 'tell' but it isn't that easy. I hacked about the
> dvbscan util to tell me more info about the streams it doesn't know about
> (I can send you a patch). From what I have seen yeah the audio streams
> have an 0x0A tag on them and they are easy to recognize. But the both the
> Audio and Video streams have the 0x52 tag on them (the audio usually has
> the 0x0A first though). So the patch seems correct unless they change the
> order of these tags.

The audio stream tag (0x0A) is also looked at the next switch statement
and used to parse the language code. This is why I asked on the list a
few days ago about if the subtitle streams etc are always in a private
section. If they were then the whole lot could be consolidated into a
much nicer bit of logic.

Unfortunately it seems that things were done that way because it is
not guaranteed that subtitle streams etc are always in private sections

:-(

>
> Stuart Auchterlonie wrote:
> >While we are talking about major TODOs there will be a need for a
> >mechanism for the frontend to tell the backend which PIDs it wants.
>
> I think this would be much simpler done if the backend simply records all
> streams given in the channel description. The front end can then pick and
> choose which streams to play. This will allows you to record interactive
> content, teletext, subtitles... and then maybe later when viewing select
> the stream you want to listen/watch though the interactive interface or a
> manually cycle through all (suspected) audio/video streams.

What about when you need to pick a program off a different transport
stream?

Streams that the mheg application wants are coded in the form
dvb://<original_network_id>.[<transport_stream_id>].<service_id>

It is theoretically possible that a different transport is being used
is it not????


Stuart


rtjacob at earthlink

May 5, 2005, 9:51 AM

Post #10 of 11 (1600 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

Quoting Stuart Auchterlonie <stuarta[at]squashedfrog.net>:
> > >While we are talking about major TODOs there will be a need for a
> > >mechanism for the frontend to tell the backend which PIDs it wants.

> > I think this would be much simpler done if the backend simply records all
> > streams given in the channel description. The front end can then pick and
> > choose which streams to play. This will allows you to record interactive
> > content, teletext, subtitles... and then maybe later when viewing select
> > the stream you want to listen/watch though the interactive interface or a
> > manually cycle through all (suspected) audio/video streams.

> What about when you need to pick a program off a different transport
> stream?

Sounds like an issue that will make InteractiveTV only really usable when using
livetv and caught up.. I never really understood how one would use mheg from a
recording with the type of setup you are talking about over there in the UK
unless you recording the entire transport stream but that sure would eat up hdd
space quick..

> Streams that the mheg application wants are coded in the form
> dvb://<original_network_id>.[<transport_stream_id>].<service_id>

> It is theoretically possible that a different transport is being used
> is it not????



_______________________________________________
mythtv-dev mailing list
mythtv-dev[at]mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev


stuarta at squashedfrog

May 5, 2005, 10:18 AM

Post #11 of 11 (1623 views)
Permalink
Re: Re: [mythtv-commits] mythtv siparser changes.... [In reply to]

On Thu, May 05, 2005 at 12:51:33PM -0400, Taylor Jacob wrote:
>
> Sounds like an issue that will make InteractiveTV only really usable when using
> livetv and caught up.. I never really understood how one would use mheg from a
> recording with the type of setup you are talking about over there in the UK
> unless you recording the entire transport stream but that sure would eat up hdd
> space quick..
>

I don't think it's sensible to have interactive tv from a recording.

But near realtime would work okay. Much of the content is cached
as it is objects (png graphics, mheg apps etc). When delayed viewing
would fail is when the realtime streams aren't available.

The target for this would be live tv (or slightly delayed) but not
recorded and watched at a later date.


Stuart

MythTV dev RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact lists@gossamer-threads.com
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.