Me TV (a GStreamer/GTK DVB-T player) was working fine on my Debian Sid
laptops 2017-08-24. I will admit to using gstreamer1* packages from
experimental, but maybe I should give this up and revert to using
I started making some changes to the Me TV UI but not to the GStreamer
(just playbin really) stuff as far as I know. I got to the stage of
doubting everything so pulled the known to work 2017-08-24 executable
from my backups, I also got the definitely known to work 2017-07-23
executable. All of these give the same problem when trying to play a
DVB stream, which indicates that the problem is actually the gstreamer
packages and plugins on my system.
The error message I am currently getting in all situations is:
terminate called after throwing an instance of 'std::runtime_error'
what(): adapter0, frontend0: GStreamer error: state change failed and some element failed to post a proper error message with the reason for the failure.
which results from an exception I throw in my GStreamer bus listener on
receiving a GST_ERROR_MESSAGE on the bus.
Given the message is immediate and nothing else GStreamer-ish has
happened, I am guessing an initialisation issue. The channel file is
present and works with dvbv5-zap/mplayer.
On Mon, 2017-09-04 at 14:25 +0100, Charlie Turner wrote:
> Hi Russel,
> The first step would be exporting GST_DEBUG before running your
> to a suitable value to get more information about what is failing.
> export GST_DEBUG=4 isn't a bad place to start, you can go up in
> if it's still not giving you anything interesting.
> The docs for GST_DEBUG are here,
> https://gstreamer.freedesktop.org/data/doc/gstreamer/head/gstreamer/h > tml/gst-running.html
Thanks for the email and reminding me the number of times Sebastian has
told me to do this (though he suggested level 3). Clearly I need
reminders of this on a regular basis! Anyway, this has been hugely
successful – the GStreamer issue was it was trying to start a GL
pipeline when such a pipeline is inappropriate on this hardware, and
indeed fails (*). Why this has come up now is to do with a Debian
upgrade, and a change in the driver set, coupled with a change those
changes make in the way this laptop reboots. I already had a --no-gl
option to force not using a GL pipeline, but I haven't had to use it on
this hardware before.
So now I have stuff working, it is on to the more enjoyable stuff of
finding code to deal with the new situation so the explicit option is
And then back to the head banging of trying to get EPG data. :-)
Thanks again for the impetus to get past the "brick wall". Much
(*) I am going to have to find a nicer way of terminating for the final