xine - a free video player

hi bastien,

> > but seriously: it's just like i said - the xine-engine has a stable api 
> > now, so there's no reason to have one specific "official" frontend from 
> > now on. users are free to choose which frontend they want to use. by the 
> > weekend i think only xine-ui and gnome-xine will work with the new api, 
> 
> Oh really ? I already ported Totem to the new API.

hey - cool job :)

> Although I'm having 2 problems. First, gtk+ lockups (I'm happy, the
> "official" front-end is broken too ;).

humm - not sure what you're talking about, but at least gnome-xine
doesn't lock up since i removed a deadlock 5 minutes ago miguel introduced ;>

so, please check out the latest cvs and try again. if you still have trouble
with deadlocks, could you please try to describe what stream type you're
trying to play back and how you provoke the deadlock (of course, a stack trace
of all threads would be the best, but might be too much asked ;))


> And I'm having troubles saving all the entries of the configuration. It
> only saves those few config entries I modify.

humm - make sure you register all config items - config items that
are not registered are dropped (so the config file gets cleaned of old
no-longer used entries).

cheers,

   guenter

--
      "Education is what remains after one has forgotten everything he 
learned in school." - Albert Einstein

Hi guenter,

> > stress the metronom. The patch for the new metronom is attached for
> > review and if noone shouts STOP I would like to commit it today or
> > tomorrow.
>
> oh dear - why all of a sudden everybody feels the need to tweak
> metronom again ?!?
>
> it was working so perfectly and there are hundrets of hours of work
> in it.

Well, almost perfect. I forgot to explain the reason for the patch:
Thibaut was not happy with the seeking. (And since I want to write a 
video editing app based on xine, scratch seeking would be really 
useful.) I was not happy with some discontinuities on DVDs, especially 
multiangle discs have a small audio gap when the angles start or stop.
All the other changes to discontinuities basically just end the 
discontinuity window earlier, which makes short streams like dvd menu 
transitions more fluent, because metronom would use the new PTS values 
earlier.

> i don't want to stop you - but i must confess that i don't really
> want to get into metronom testing again - and i don't really have the
> time for that either. just be carefull with what you do :-)

I will revert the patch as soon as someone shows me a stream that worked 
better with the old metronom and I fail to fix the new one. Is that ok 
for you? Or would you feel more comfortable if I wait until you have 
your exams done? (We all have an off-xine life that requires some 
attention every now and then... ;) Although getting this into cvs would 
make developing easier for me I could try to maintain it separately.

> > But let my explain this thing a bit:
>
> [...]
>
> from your explanation everything sounds reasonable, i just hope the
> code really works. my impression is that most of your changes deal
> with discontinuities and fifo flushes - so they shouldn't affect any
> of the metronom stress test streams which contain broken pts values
> (mostly mpeg-1 files) since they do not have discontinuities.

Sounds good. After all the changes are less drastical then they look. 
The more difficult trial&error parts of metronom (video drift & PTS 
interpolation) are not touched. It is basically just a more 
sophisticated discontinuity handling, output fifo flushing and seek 
improvement by waiting for at least one frame.

Michael

-- 
printk(KERN_WARNING "Multi-volume CD somehow got mounted.\n");
	2.2.16 /usr/src/linux/fs/isofs/inode.c

Hi guenter,

> > 4) Multiple Subtitles displayed at the same time. DVD spus and CC
> > working together.
> > (Decide on some sort of xine-lib api so the GUI can set the second
> > subtitle channel.)
>
> more and more people seem to request that the whole subtitle c**p
> should get an extra xine-api call, this would also stop the trouble
> with ever-changing subtitle file delimiters (...or rather turn that
> into a frontend problem ;))

I maintain the following oppinion on that:
- one spu decoder per stream
- if multiple spus shall be displayed, multiple streams should be used
- the text subtitles should be turned into such and additional stream
This is what the multi-stream feature is for and I think this is the 
cleanest (no new workarounds) and most flexible way.

The feature (hack) of text subtitles added to a MRL with some delimiter 
should be removed if this works. You would have two streams started 
with two MRLs then: One for the main video and one for the text 
subtitle.

Michael

-- 
"A computer programmer is someone who, when told to "Go to Hell", sees
the "Go to", rather than the destination, as harmful."

Hi James,

> >>4) Multiple Subtitles displayed at the same time.
> >>DVD spus and CC working together.
> >>(Decide on some sort of xine-lib api so the GUI can set the second
> >>subtitle channel.)
> >
> >I don't think we really need multiple spu plugins for this
> > application. Since the CC is a side product of libmpeg2, I would
> > suggest taking this completely out of the normal spu decoder chain
> > and just call the decode_data() from the CC decoder from within
> > libmpeg2.
>
> The intention is to move the CC decoder to it's own decoder plugin,
> for the purpose of future TV capture support etc. So if it is a
> separate decoder, we cannot call decode_data until the plugin is
> loaded, and thus the problem of needing multiple spu decoder plugins.

We still can. Just instantiate the plugin from within libmpeg2. You have 
access to the plugin catalog, so you can call open_plugin on the CC 
decoder.

For other corner cases we can always use additional streams for 
additional spu channels.

> >>It would be nice to find a method to match buf_type to spu_decoder
> >>without having to search the entire list of spu_decoders each time
> >> a fifo buf comes.
> >
> >Is already done by the plugin loader. There is a mapping of buffer
> > type to decoder plugin that is done once at plugin loading and is
> > then used to quickly find the right decoder.
>
> It is just a big switch-case statement in "void *video_decoder_loop
> (void *stream_gen)"
> as far as I can tell at the moment.

get_spu_decoder() in load_plugins.c is called from there and this one 
uses the decoder map.

Michael

-- 
/* Nobody will ever see this message :-) */
panic("Cannot initialize video hardware\n");
	2.0.38 /usr/src/linux/arch/m68k/atari/atafb.c

hi michael,

> > i don't want to stop you - but i must confess that i don't really
> > want to get into metronom testing again - and i don't really have the
> > time for that either. just be carefull with what you do :-)
> 
> I will revert the patch as soon as someone shows me a stream that worked 
> better with the old metronom and I fail to fix the new one. Is that ok 
> for you?

sounds good to me - as i said, i don't want to stop you as everything
you intend to do sounds very reasonable. i don't have any multiangle dvds
(i think) so i wasn't aware of any problems in that area.

keep up the good work,

   guenter

--
"Plaese porrf raed."
                -- Prof. Michael O'Longhlin, S.U.N.Y. Purchase

On Mon, 2002-10-28 at 20:34, Guenter Bartsch wrote:
> hi bastien,
>=20
> > > but seriously: it's just like i said - the xine-engine has a stable a=
pi=20
> > > now, so there's no reason to have one specific "official" frontend fr=
om=20
> > > now on. users are free to choose which frontend they want to use. by =
the=20
> > > weekend i think only xine-ui and gnome-xine will work with the new ap=
i,=20
> >=20
> > Oh really ? I already ported Totem to the new API.
>=20
> hey - cool job :)

Tada :)

> > Although I'm having 2 problems. First, gtk+ lockups (I'm happy, the
> > "official" front-end is broken too ;).
>=20
> humm - not sure what you're talking about, but at least gnome-xine
> doesn't lock up since i removed a deadlock 5 minutes ago miguel introduce=
d ;>

That fixed it (after I fixed a bug in my mouse events generation, and a
problem with int v. double for the display_ratio)

> so, please check out the latest cvs and try again. if you still have trou=
ble
> with deadlocks, could you please try to describe what stream type you're
> trying to play back and how you provoke the deadlock (of course, a stack =
trace
> of all threads would be the best, but might be too much asked ;))

Cracked it :)

> > And I'm having troubles saving all the entries of the configuration. It
> > only saves those few config entries I modify.
>=20
> humm - make sure you register all config items - config items that
> are not registered are dropped (so the config file gets cleaned of old
> no-longer used entries).

Hmm, how would I go around doing that ? I'm not sure I understand the
config system in Xine...

Cheers

--=20
/Bastien Nocera
http://hadess.net

Perfection is reached, not when there is no longer anything to add, but
when
there is no longer anything to take away.
						Antoine de Saint-Exupery

hi bastien,

> > > And I'm having troubles saving all the entries of the configuration. It
> > > only saves those few config entries I modify.
> > 
> > humm - make sure you register all config items - config items that
> > are not registered are dropped (so the config file gets cleaned of old
> > no-longer used entries).
> 
> Hmm, how would I go around doing that ? I'm not sure I understand the
> config system in Xine...

it's xine, not Xine

the config system is very simple: the config file gets loaded (though 
you could also use your own mechanism here and store the config values 
using gconf) and all entries found in the file are added to xine's 
internal config repository. all config entries are marked as 
unregistered (unclaimed) at that point. then modules can start calling 
xine_config_register_* functions to register their config values, 
providing data types, descriptions and default values (among other 
things). xine_config will then check if an unregistered item with that 
key exists and if that is the case, convert into an registered entry and 
return the value that was loaded from the configfile. if the key is not 
found, a new entry will be added to the config repository and the 
default value will be copied and also returned.

when you call xine_config_save (...) xine_config will iterate through 
the list of config items and save those that have been registered in the 
meantime and discard everything else.

cheers,

   guenter

--
Math is like love -- a simple idea but it can get complicated.
                -- R. Drabek

Michael Roitzsch wrote:

>Hi guenter,
>
>  
>
>>>4) Multiple Subtitles displayed at the same time. DVD spus and CC
>>>working together.
>>>(Decide on some sort of xine-lib api so the GUI can set the second
>>>subtitle channel.)
>>>      
>>>
>>more and more people seem to request that the whole subtitle c**p
>>should get an extra xine-api call, this would also stop the trouble
>>with ever-changing subtitle file delimiters (...or rather turn that
>>into a frontend problem ;))
>>    
>>
>
>I maintain the following oppinion on that:
>- one spu decoder per stream
>- if multiple spus shall be displayed, multiple streams should be used
>- the text subtitles should be turned into such and additional stream
>This is what the multi-stream feature is for and I think this is the 
>cleanest (no new workarounds) and most flexible way.
>
>The feature (hack) of text subtitles added to a MRL with some delimiter 
>should be removed if this works. You would have two streams started 
>with two MRLs then: One for the main video and one for the text 
>subtitle.
>
>Michael
>
>  
>
I thought that a new stream would have it's own independant metronom 
clock, so if we put the spu decoder in a stream on it's own, how will it 
sync up with the other stream ?

Cheers
James

Hi James,

> >I maintain the following oppinion on that:
> >- one spu decoder per stream
> >- if multiple spus shall be displayed, multiple streams should be
> > used - the text subtitles should be turned into such and additional
> > stream This is what the multi-stream feature is for and I think
> > this is the cleanest (no new workarounds) and most flexible way.
> >
> >The feature (hack) of text subtitles added to a MRL with some
> > delimiter should be removed if this works. You would have two
> > streams started with two MRLs then: One for the main video and one
> > for the text subtitle.
>
> I thought that a new stream would have it's own independant metronom
> clock, so if we put the spu decoder in a stream on it's own, how will
> it sync up with the other stream ?

Yes this is true for now, but Siggi already mentioned in an earlier mail 
that we need the possibility to sync multiple streams against a common 
clock. I think when this 1.0 prerelease has settled and the 
architecture stabilized we can start brainstorming some extensions to 
metronom. Something like a global part of metronom with one clock for 
all streams.

Michael

-- 
panic("Lucy in the sky....");
	2.2.16 /usr/src/linux/arch/sparc64/kernel/starfire.c

Hi guenter,

> when you call xine_config_save (...) xine_config will iterate through
> the list of config items and save those that have been registered in
> the meantime and discard everything else.

How is that supposed to work? I start xine, play, say, an avi and stop 
xine. This way I lose all config entries related to liba52 because it 
never got loaded and never had a chance to register its config items.

Michael

-- 
"Your job is being a professor and researcher: That's one hell of a
 good excuse for some of the brain-damages of minix."
	-Linus Torvalds to Andrew Tanenbaum

Hi guenter,

> > > i don't want to stop you - but i must confess that i don't really
> > > want to get into metronom testing again - and i don't really have
> > > the time for that either. just be carefull with what you do :-)
> >
> > I will revert the patch as soon as someone shows me a stream that
> > worked better with the old metronom and I fail to fix the new one.
> > Is that ok for you?
>
> sounds good to me - as i said, i don't want to stop you as everything
> you intend to do sounds very reasonable. i don't have any multiangle
> dvds (i think) so i wasn't aware of any problems in that area.

It is in cvs now, so anyone with a bit of time and badly behaving 
streams at hand: give the new (and hopefully improved) metronom a try.

Michael

-- 
panic("IRQ, you lose...");
	2.2.16 /usr/src/linux/arch/mips/sgi/kernel/indy_int.c

Am Dienstag, 29. Oktober 2002 03:08 schrieb Guenter Bartsch:

> it's xine, not Xine

Shouldn't this be changed in the xine.desktop file, there it is called 
'Xine'. Everyone can read Xine in the KDE K-Menu (when using KDE) this 
will not force the right spelling.

-- 
Machs gut    | http://www.iivs.de/schwinde/buerger/tremmel/

Manfred      | http://www.knightsoft-net.de

hi manfred,

> > it's xine, not Xine
> 
> Shouldn't this be changed in the xine.desktop file

actually i'm still not sure what a .desktop exactly is (still didn't 
have time to really dig into either gnome or kde - yes, i know this is 
bad as i'm the author of gnome-xine) but it definitely should be fixed

... daniel? :)

cheers,

   guenter

--
We're only in it for the volume.
                -- Black Sabbath

Hi Guenter,

Guenter Bartsch wrote:
> hi manfred,
> 
> 
>>>it's xine, not Xine
>>
>>Shouldn't this be changed in the xine.desktop file
> 
> 
> actually i'm still not sure what a .desktop exactly is (still didn't 
> have time to really dig into either gnome or kde - yes, i know this is 
> bad as i'm the author of gnome-xine) but it definitely should be fixed
> 
> .... daniel? :)

	Done.

Cheers.
-- 
73's de Daniel, F1RMB.

              -=- Daniel Caujolle-Bert -=- segfault@... -=-
                        -=- f1rmb@... (AMPR NET) -=-

Am Dienstag, 29. Oktober 2002 22:21 schrieb Guenter Bartsch:

> actually i'm still not sure what a .desktop exactly is (still didn't
> have time to really dig into either gnome or kde - yes, i know this
> is bad as i'm the author of gnome-xine) but it definitely should be
> fixed

I don't have time today, but if you want, I will send you tomorow a 
fixed version.

-- 
Machs gut    | http://www.iivs.de/schwinde/buerger/tremmel/

Manfred      | http://www.knightsoft-net.de

On Tue, 2002-10-29 at 02:08, Guenter Bartsch wrote:
> hi bastien,
>=20
> > > > And I'm having troubles saving all the entries of the configuration=
. It
> > > > only saves those few config entries I modify.
> > >=20
> > > humm - make sure you register all config items - config items that
> > > are not registered are dropped (so the config file gets cleaned of ol=
d
> > > no-longer used entries).
> >=20
> > Hmm, how would I go around doing that ? I'm not sure I understand the
> > config system in Xine...
>=20
> it's xine, not Xine

Tomato, tomato...

> the config system is very simple: the config file gets loaded (though=20
> you could also use your own mechanism here and store the config values=20
> using gconf) and all entries found in the file are added to xine's=20
> internal config repository. all config entries are marked as=20
> unregistered (unclaimed) at that point. then modules can start calling=20
> xine_config_register_* functions to register their config values,=20
> providing data types, descriptions and default values (among other=20
> things). xine_config will then check if an unregistered item with that=20
> key exists and if that is the case, convert into an registered entry and=20
> return the value that was loaded from the configfile. if the key is not=20
> found, a new entry will be added to the config repository and the=20
> default value will be copied and also returned.
>=20
> when you call xine_config_save (...) xine_config will iterate through=20
> the list of config items and save those that have been registered in the=20
> meantime and discard everything else.

Well, I was trying to figure out why it wouldn't work for totem, but I
found out. Libgnomeui was saving the placement for some widgets in the
same place I was saving and loading my configuration. It's working now
though, so Totem is very well on track for a release along with the new
X^Wxine.

But where is the logo gone ? ;)

Cheers

--=20
/Bastien Nocera
http://hadess.net

Perfection is reached, not when there is no longer anything to add, but
when
there is no longer anything to take away.
						Antoine de Saint-Exupery

hi bastien,

On 10/30, Bastien Nocera wrote:

> But where is the logo gone ? ;)

...humm, i guess the xine engine got frustrated over all those frontends 
replacing the logo (often by logos that didn't even contain the name 
"xine" in any way) so it stopped displaying the logo. frontends that 
want a logo can generate their own small logo stream and display it when 
they're idle - see gnome-xine for an example (playlist.c:play_logo)

cheers,

   guenter

--
There is no TRUTH.  There is no REALITY.  There is no CONSISTENCY.
There are no ABSOLUTE STATEMENTS   I'm very probably wrong.

On 30 Oct 2002, Bastien Nocera wrote:

[...]
> But where is the logo gone ? ;)

Some frontends seemed to prefer an own logo, and they made big efforts to
work arounnd xine's internal logo, so we decided to off-load the logo
display code to frontends.

Cheers,
	Siggi

On Wed, 2002-10-30 at 11:18, Guenter Bartsch wrote:
> hi bastien,
> 
> On 10/30, Bastien Nocera wrote:
> 
> > But where is the logo gone ? ;)
> 
> ...humm, i guess the xine engine got frustrated over all those frontends 
> replacing the logo (often by logos that didn't even contain the name 
> "xine" in any way) so it stopped displaying the logo. frontends that 

Right. If you wanted fame, you should have used a license with
advertising clause...

> want a logo can generate their own small logo stream and display it when 
> they're idle - see gnome-xine for an example (playlist.c:play_logo)

Will look at it.

Thanks

-- 
/Bastien Nocera
http://hadess.net

Perfection is reached, not when there is no longer anything to add, but
when
there is no longer anything to take away.
						Antoine de Saint-Exupery

hi bastien,

> > ...humm, i guess the xine engine got frustrated over all those frontends 
> > replacing the logo (often by logos that didn't even contain the name 
> > "xine" in any way) so it stopped displaying the logo. frontends that 
> 
> Right. If you wanted fame, you should have used a license with
> advertising clause...

...so you believe that giving proper credit is only appropriate if the 
license enforces it?

   guenter

On Fri, 2002-11-01 at 12:04, Guenter Bartsch wrote:
> > Right. If you wanted fame, you should have used a license with
> > advertising clause...
> 
> ...so you believe that giving proper credit is only appropriate if the 
> license enforces it?

ok, we can try to make better... instead of license with advertising
clause we can create a small banner image "powered by xine" and add to
package. documentation should just say that "it would be nice to use".

regards,

Miguel

On Fri, 2002-11-01 at 14:04, Guenter Bartsch wrote:
> hi bastien,
> 
> > > ...humm, i guess the xine engine got frustrated over all those frontends 
> > > replacing the logo (often by logos that didn't even contain the name 
> > > "xine" in any way) so it stopped displaying the logo. frontends that 
> > 
> > Right. If you wanted fame, you should have used a license with
> > advertising clause...
> 
> ...so you believe that giving proper credit is only appropriate if the 
> license enforces it?

No, I'm just saying that in Totem, your copyrights and credits are in there.
If you feel that me changing the logo isn't appropriate and you want the
xine name to be more user-visible, I can do that.

Removing the logo feature from the engine was one way to "enforce" that,
but it's not very nice to front-ends developers.

Regards

-- 
/Bastien Nocera
http://hadess.net

Perfection is reached, not when there is no longer anything to add, but
when
there is no longer anything to take away.
						Antoine de Saint-Exupery

hi bastien,

On 11/01, Bastien Nocera wrote:

> > > Right. If you wanted fame, you should have used a license with
> > > advertising clause...
> > 
> > ...so you believe that giving proper credit is only appropriate if the 
> > license enforces it?
> 
> No, I'm just saying that in Totem, your copyrights and credits are in there.
> If you feel that me changing the logo isn't appropriate and you want the
> xine name to be more user-visible, I can do that.

the final decision is up to you, of course :) 

> Removing the logo feature from the engine was one way to "enforce" that,

huh? giving frontends more freedom is enforcing them to do something?!?

actually the logo (and the automatically generated OSD) was removed for 
one reason: advanced frontends (e.g. audio/video processing 
applications) would get pretty annoyed by these - plus it doesn't really 
make sense with the streams api. Later it also turned out that by 
removing the logo from the engine a lot of hacks and tweak where no 
longer needed so the engine code gets cleaner and hopefully more stable.

> but it's not very nice to front-ends developers.

it's 3 lines more for playback-only frontends (but those have more 
flexibility in return, e.g. can choose to display different logos 
depending on the skin selected etc.) and -x0000 lines of tweaking code 
for advanced frontends.

cheers,

   guenter

--
Real programmers don't comment their code.  It was hard to write, it
should be hard to understand.

On Fri, 2002-11-01 at 16:46, Guenter Bartsch wrote:
> hi bastien,
> 
> On 11/01, Bastien Nocera wrote:
> 
> > > > Right. If you wanted fame, you should have used a license with
> > > > advertising clause...
> > > 
> > > ...so you believe that giving proper credit is only appropriate if the 
> > > license enforces it?
> > 
> > No, I'm just saying that in Totem, your copyrights and credits are in there.
> > If you feel that me changing the logo isn't appropriate and you want the
> > xine name to be more user-visible, I can do that.
> 
> the final decision is up to you, of course :) 
> 
> > Removing the logo feature from the engine was one way to "enforce" that,
> 
> huh? giving frontends more freedom is enforcing them to do something?!?
> 
> actually the logo (and the automatically generated OSD) was removed for 
> one reason: advanced frontends (e.g. audio/video processing 
> applications) would get pretty annoyed by these - plus it doesn't really 
> make sense with the streams api. Later it also turned out that by 
> removing the logo from the engine a lot of hacks and tweak where no 
> longer needed so the engine code gets cleaner and hopefully more stable.
> 
> > but it's not very nice to front-ends developers.
> 
> it's 3 lines more for playback-only frontends (but those have more 
> flexibility in return, e.g. can choose to display different logos 
> depending on the skin selected etc.) and -x0000 lines of tweaking code 
> for advanced frontends.

Sorry, I misunderstood you. The way you put it, I saw a cause to
consequence behaviour. If it wasn't for that reason, then please accept
my apologies. Be clearer as to *why* you did something next time ;)

Cheers

-- 
/Bastien Nocera
http://hadess.net

Perfection is reached, not when there is no longer anything to add, but
when
there is no longer anything to take away.
						Antoine de Saint-Exupery