2520 replies to this topic

[christophecvr]

Ter info gstreamer recipes for 1.9.0.1 updated to today 2013-05-05 all ok.

 

https://github.com/c...TREAMER-RECIPES

I mean    2016-05-05

Edited by christophecvr, 5 May 2016 - 16:34.

[MastaG]

Hi Chris,

They've changed the adaptivedemux code again in gst-plugins-bad: https://cgit.freedes...0558b5be2f01715

So the patch doesn't apply anymore, maybe patch is not required anymore.

Anyways, I'm just letting you know since I'm going take a break for a while

[christophecvr]

Hi Chris,

They've changed the adaptivedemux code again in gst-plugins-bad: https://cgit.freedes...0558b5be2f01715

So the patch doesn't apply anymore, maybe patch is not required anymore.

Anyways, I'm just letting you know since I'm going take a break for a while

Thank's for informing I checked the bug report about this issue even yesterday morning but then it was not pushed yet. In the mean time well. I'll adapt the recipes

[christophecvr]

Gstreamer 1.9.0.1 recipes update on :

 

https://github.com/c...TREAMER-RECIPES

[christophecvr]

@all self builders.

 

gues all now that with some media especially that with has been re rendered to have a smaller filesize that when pasing to paused. The media blocked.

We where unable to relaunch or unpause the media only stoping the media helped.

For some reason gstreamer lost info about audio track in use. This resulted into a pipeline stuck into state playing pending state paused.

This issue also blocked all further options for the media in use . The only option was to stop media.

This patch solves this issue.

 

See further comments in the patch included here.

 

I well set into the patch a certain very good procedure of mx3L which was limited to gst-1.0 also avbl for gst-0.10.

According to manual off gst-0.10 it should work. But this will need testing by a self builder who still build his image pli4 with gst-0.10. to be shure.

I'm almost shure it will be ok and a very nice improvement for the still going on beta pli4 image with gst-0.10.

 

A typical media whit such pause/unpause problems I know it's a very ugly one but nice for test.

 

https://www.dropbox....ldlife.avi?dl=1

 

 

Attached Files

•  E2-serviceMP3-pause-unpause-issue.patch   4.27KB   43 downloads

[littlesat]

Merged in next-master....

[christophecvr]

Merged in next-master....

Nice but like MastaG builds with master (I also now) this cause the next-master was not always kept in line with master.

 

But I saw the patch was downloaded many times without any report (negatif or positif) . Also it would be nice if a person who build self and uses pli4 standard (which means the use of very deprecated gst-0.xx) would test it. I'm almost shure it will all work on this very very old and deprecated gstreamer version. The advantage is that we can reamein in line with e2 master if tests are ok.

[mx3L]

I well set into the patch a certain very good procedure of mx3L which was limited to gst-1.0 also avbl for gst-0.10.

According to manual off gst-0.10 it should work

Maybe there is a reason why I did that:

1. I don't post patches which are not tested

2. souphttpsrc will timeout in 0.10

[christophecvr]

 

I well set into the patch a certain very good procedure of mx3L which was limited to gst-1.0 also avbl for gst-0.10.

According to manual off gst-0.10 it should work

Maybe there is a reason why I did that:

1. I don't post patches which are not tested

2. souphttpsrc will timeout in 0.10

 

Nice to know , ... The question I did asked more then once and it's the first time you respond in a correct way to it ????

 

OK to all who are trying with this is even confirmed it should work 100 % ok for all file media.

It might (according to mx3L but I gues he's really right on this mather) cause problems for http media on 0.10.

 

Well no problem as it can be adapted for gst-1.0 file media and souphttp media.

For gst-0.10 only file media .

[mx3L]

@littlesat is also part of the discussion, he can also tell how many times did you ask and what you asked. I told you it's ment only for gstreamer1.0 you didn't ask why.

I'm not making stories, why would I?

 

One last thing.

I would suggest Openpli to make sure that patches are understood before they're pushed, since I can see it's not happening in all cases. If you are not sure and you want to push them then check if author's reasoning makes sense and if author provides enough evidence to support his claims(logs, samples, how to reproduce problem, on what receiver(s) it was tested and if testing was done in relevant conditions, i.e. for example no custom sink was used if it can change results of tests, or at least author mentions it).

[christophecvr]

@littlesat is also part of the discussion, he can also tell how many times did you ask and what you asked. I told you it's ment only for gstreamer1.0 you didn't ask why.

I'm not making stories, why would I?

 

One last thing.

I would suggest Openpli to make sure that patches are understood before they're pushed, since I can see it's not happening in all cases. If you are not sure and you want to push them then check if author's reasoning makes sense and if author provides enough evidence to support his claims(logs, samples, how to reproduce problem, on what receiver(s) it was tested and if testing was done in relevant conditions, i.e. for example no custom sink was used if it can change results of tests, or at least author mentions it).

Nice to see you're extreme positive attitude .

 

But did a request to push this last patch right away into master ?? no

I just expressed that and that is trough concerning next-master off e2 is used by no one.

 

For the rest the patch is needed to be tested by everynone who is willing constructif remarks are really welcome taht's all.

[mx3L]

I'm not talking only about this patch, but I resigned to comment on everything I don't agree with, since most of the time it had no result or it was only slightly altered. And reading and answering to unnecessary long posts with hard to understand English began to be tiresome.

[christophecvr]

Nice talk mx3L but ..... ????

 

@ all who want to try.

 

1) on gst-1.0 all ok (please give test reports)

 

2) On gst-0.10 I would have been tempted adapt the patch a little so that only file media would be concerned. But sorry after the comments without any prove or sense from mx3L

     It would be nice to test all anyway on gst-0.10 http media and see if he's right ??? about the time out  . he's to aggressive with no reason about the fact it does not work on gst-0.10.

     The reaction is out off bound and makes no sense. If it was really like that mx3L would have told me long before in a descent way like he always did before. Now not and that's .... sorry .....

[hemertje]

Dont start again fighting...

Thank you!

[dazulrich]

@ all,

 

apologies for resurecting this thread, but the question fits this best.

Frist first, this is coming from a non openpli image, using Christophe's multi-sink. (but it could well be relevant to the openpli sink..)

 

In another forum a couple of problematic streams were posted.

Some without sound, e.g:

http://185.38.12.34/ivs/8f/c8/0b87e5693629.mp4/hls/tracks-1%2C4/index.m3u8

When looking into the gst logs, its treated as AAC-RAW, and the audiosink hits this line ( which seems to date back to schwerkrafts sink)

gint rate, channels, rate_idx = 0, obj_type = 1; // hardcoded yet.. hopefully this works every time ;)

obj_type = 1 is used as profile. (which is the AAC Main profile, right?)

 

But if I understand it correctly, we only pass LC profiles to the box for decoding. (the rest would be decoded in software , if say libav was installed ?) - so some boxes with libav would always get audio.

When changing it to "obj_type = 2" , (which is the AAC LC profile) I got audio.

Should he hardcoded value be changed, so AAC_RAW always uses LC profile, if that is what the boxes support? Or pass in MAIN, and change the caps to include MAIN ?

 

there were also some streams that where stuttering on some boxes, but not on others (using Christophe's multi-box sink). E.g.

http://s1.kodik.cc/content/stream/films/b3l0m.2016.p.webdl.720p/hls/360/index.m3u8

For those streams, adding the removed use_dts flag in the videosink made all the difference.

The flag is still included in the openpli gst 1.0 sink , so I'm wondering whether this flag is indeed still useful for some streams and should be added back into Chistophe's sink?

 

thanks for reading.

 

/daz

Edited by dazulrich, 21 November 2016 - 11:37.

[christophecvr]

Try this sink just updated

 

https://github.com/c...ee/experimental

[dazulrich]

Thanks, I did just that in local repo and it worked for the boxes I have. but that is a very small sample.. :-)

[foxbob]

Not sound in play dts.wav file:

 

root@vusolo2:~# gst-launch-1.0 playbin uri=file:///media/hdd/music/Bad.wav[J

Setting pipeline to PAUSED ...

Pipeline is PREROLLING ...

WARNING: from element /GstPlayBin:playbin0/GstPlaySink:playsink: No volume control found

Additional debug info:

../../../git/gst/playback/gstplaysink.c(2883): gen_audio_chain (): /GstPlayBin:playbin0/GstPlaySink:playsink:

Volume/mute is not available

Pipeline is PREROLLED ...

Setting pipeline to PLAYING ...

New clock: GstSystemClock

ERROR: from element /GstPlayBin:playbin0/GstPlaySink:playsink/GstBin:abin/GstDVBAudioSink:dvbaudiosink0: Could not determine type of stream.

Additional debug info:

../git/gstdvbaudiosink.c(791): gst_dvbaudiosink_set_caps (): /GstPlayBin:playbin0/GstPlaySink:playsink/GstBin:abin/GstDVBAudioSink:dvbaudiosink0:

hardware decoder can't be set to bypass mode type audio/x-dts

Execution ended after 0:00:00.008177148

Setting pipeline to PAUSED ...

Setting pipeline to READY ...

Setting pipeline to NULL ...

Freeing pipeline ...

[Beeker]

GStreamer GIT master head of today big problem.

 

First a small one.

 

GStreamer-plugins-bad fails to compile due to uninitialized variable.

 

Ok that easy.

diff --git a/sys/decklink/gstdecklink.cpp b/sys/decklink/gstdecklink.cpp
index 955157b..1fe9005 100644
--- a/sys/decklink/gstdecklink.cpp
+++ b/sys/decklink/gstdecklink.cpp
@@ -736,7 +736,7 @@ public:
         GstClockTime packet_time, gboolean no_signal) = NULL;
     GstDecklinkModeEnum mode;
     GstClockTime capture_time = GST_CLOCK_TIME_NONE;
-    GstClockTime base_time;
+    GstClockTime base_time = 0;
     gboolean no_signal = FALSE;
     GstClock *clock = NULL;
     HRESULT res;

But gstreamer-base breaks cd, mp's and wav.

 

A high pitch noise due to this

 

https://cgit.freedes...6f149b3b023882d

 

I  reported that to gstreamer-dev's, but gstreamer-dev ask me to make a new threat.I will do that this evening.

 

So don't upgrade to master HEAD unless you want to get rid of your sound speakers.

Edited by Beeker, 29 November 2016 - 17:55.

[MastaG]

There's always uninitialized variables that will break compiling lol.

Thank you Beeker for the bug report.