RSB GStreamer Integration: Issueshttps://code.cor-lab.de/https://code.cor-lab.de/favicon.ico?14019720732015-03-20T14:53:30ZOpen Source Collaboration Platform
Redmine Bug #2207 (New): create timestamps in videosink are wronghttps://code.cor-lab.de/issues/22072015-03-20T14:53:30ZJ. Wienkejwienke@techfak.uni-bielefeld.de
<pre>
Event
Scope : /foo/
Id : 7BA80410-3F10-5FC0-A517-319754D5FC91
Sequence-Number: 126
Origin : 9257D36D-1076-4153-8E4D-D0E5AB6D393D
Method : N/A
Timestamps
Create : 2015-03-17T11:50:49.350336+01:00
Send : 2015-03-20T15:52:15.707611+01:00
Receive: 2015-03-20T15:52:15.716142+01:00
Deliver: 2015-03-20T15:52:15.716151+01:00
</pre> Bug #2067 (New): Fails on ambiguous .rst.vision.Image converter definitionhttps://code.cor-lab.de/issues/20672014-10-29T15:04:41ZC. Leichsenringcmertes@cit-ec.uni-bielefeld.de
<p>Ambiguous configurations of the .rst.vision.Image cause any rsb-gstreamer component to fail with a message along these lines:<br /><pre>
0:00:00.098899690 14722 0x1614a00 ERROR rsbsrc /media/local_ci/jenkins/jobs/rsb-gstreamer-resume_stream_fix-toolkit-nightly/workspace/label/master/src/gst_rsb_src_base.cpp:341:gst_rsb_src_base_ensure_listener:<rsbaudiosrc0> failed to create RSB listener: Error configuring connector `spread', in-push: Ambiguous converter set for wire-type `std::string' and wire-schema `.rst.vision.Image': candidate data-types are {rst::vision::Image, _IplImage}; hint: add a configuration option `transport.<name>.converter.cpp.".rst.vision.Image" = <one of {rst::vision::Image, _IplImage}>' to resolve the ambiguity (A converter is already stored for the key `.rst.vision.Image').
</pre></p>
<p>It's easy to work around this issue but since rsb-gstreamer never uses <code>_IplImage</code>, a rule to always use <code>rst::vision::Image</code> would be the more elegant behaviour.</p>
<p>From <a class="external" href="https://projects.cit-ec.uni-bielefeld.de/issues/4218">https://projects.cit-ec.uni-bielefeld.de/issues/4218</a></p> Bug #831 (New): rsbvideosink in conjunction with plugin tee fails without reasonable error messageshttps://code.cor-lab.de/issues/8312012-01-30T15:35:39ZC. Peterscpeters@cit-ec.uni-bielefeld.de
<p>I grab images from a dc1394 camera which I like to both display on the screen and publish via rsbvideosink (in order to use them with bag-record). Here's the pipeline:<br />gst-launch-0.10 dc1394src iso-speed=800 ! video/x-raw-bayer, format=gbrg, width=600, height=520, roi_left=340, roi_top=40, vmode=88, framerate=30/1 ! queue ! bayer2rgb ! ffmpegcolorspace ! tee name="splitter" ! queue ! xvimagesink splitter. ! queue ! rsbvideosink scope=/tebra/video/front</p>
<p>Logger output shows that first images are published (always 18), but subsequent images aren't. The display of images also freezes after 18 images. The pipeline doesn't quit, but just stops working properly without any error message.</p>
<p>Using either xvimagesink or rsbvideosink (without using tee) works fine:<br />gst-launch-0.10 dc1394src iso-speed=800 ! video/x-raw-bayer, format=gbrg, width=600, height=520, roi_left=340, roi_top=40, vmode=88, framerate=30/1 ! queue ! bayer2rgb ! ffmpegcolorspace ! queue ! xvimagesink<br />gst-launch-0.10 dc1394src iso-speed=800 ! video/x-raw-bayer, format=gbrg, width=600, height=520, roi_left=340, roi_top=40, vmode=88, framerate=30/1 ! queue ! bayer2rgb ! ffmpegcolorspace ! queue ! rsbvideosink scope=/tebra/video/front</p>
<p>Displaying and encoding the video in a single pipeline works fine, too:<br />gst-launch-0.10 dc1394src iso-speed=800 ! video/x-raw-bayer, format=gbrg, width=600, height=520, roi_left=340, roi_top=40, vmode=88, framerate=30/1 ! queue ! bayer2rgb ! ffmpegcolorspace ! tee name="splitter" ! queue ! xvimagesink splitter. ! queue ! theoraenc bitrate=256 ! oggmux ! filesink location="/media/local/test.ogg"</p>
<p>Hence, the problem might be the conjunction of tee and rsbvideosink, but I am not able to further specify it.</p> Bug #505 (New): More than two channels cannot be processedhttps://code.cor-lab.de/issues/5052011-08-13T04:59:24ZJ. Moringenjmoringe@cor-lab.uni-bielefeld.de
<p>When more than two channels are involved, some audio sinks (observed with ALSA and pulseaudio sinks) seems to require some sort of channel map ("channel-positions" caps item).</p>
<p>It is not clear to me how to provide this correctly. Any of my attempts lead to duplicate channel position entries which cause the sink element to freak out and the pipeline to fail.</p>
<p>Matthias, do you know how to do this?</p>
<p>This issue is rather important since Nao produces a four-channel audio stream.</p>