What am I doing to cause appsrc to hang my pipeline?

[Ben Rush]

Thanks for your kind response. I will check the response I am getting when
that function does return properly to make sure it's not an error. I will
get back to you.

And yes, I commented out the run to allow an easy way of timing out the
code. This is a rough representation of some production code that's
misbehaving (it's an attempt to get it to reproduce in a simple example for
the benefit of those answering, which it does). My production code does run
the main loop on a thread. I will update this sample code to reflect that.

On Fri, Aug 30, 2019, 3:25 PM Michael Gruner <michael.gruner at ridgerun.com>
wrote:

> Hi,
>
> I did not run your code, but at first sight it looks okay. I only have two
> comments that may be related:
>
> 1) You are not iterating on the main loop. As such, I’m surprised you even
> get signals being emitted at all. I wouldn’t rely on undefined behavior and
> iterate the loop. I imagine you commented out the g_main_loop_run to avoid
> blocking, so here’s a couple of options:
>  - call g_main_loop_run from a different thread
>  - loop yourself and call g_main_context_iteration on each iteration,
>
> 2) On the g_bus_timed_pop_filtered, are you checking the message you are
> receiving is the EOS? Because it could be an error, and you are assuming
> the EOS was received when it hasn’t.
>
>
> Michael
> www.ridgerun.com
>
>
>
> On Aug 30, 2019, at 1:48 PM, Ben Rush <ben at ben-rush.net> wrote:
>
> I have made a self-contained example here:
> https://github.com/kwende/GStreamerExamples/blob/master/Examples/RunInALoop/RunInALoop.cpp.
> The same repo includes a VS2019 solution/project to build it. I hope this
> helps someone understand my question better.
>
> Here is my problem: I think am I doing something wrong in how I'm tearing
> down the pipeline when using appsrc. If USEAPPSRC is defined in my code,
> what eventually happens is the code hangs on line 147 (after several
> iterations):
>
> gst_element_send_event(state.pipeline, gst_event_new_eos());
>
> In addition, I will sporadically get zero-length MP4 files before the hang
> eventually happens.
>
> On the other hand, if USEAPPSRC isn't defined, and I use videotestsrc,
> then everything works as planned: no zero-length files and no hangs ever
> occur: the loop continues and I keep generating files ad nauseam. I've seen
> a number of discussions about flushing the pipeline, but I'm unsure how
> this works exactly in the scenario I've got going on with this code. I feel
> like something isn't getting the EOS signal properly, but how or why, I
> can't shake out. The heart of the issue is that I'm trying to tear down and
> rebuild my pipeline in an infinite loop, I've never witnessed this issue
> unless I'm in a loop like that.
>
> It's worth noting that I'm using the Intel Media SDK h264 plugin, and so
> that might be a variable, but everything certainly works fine when using
> videotestsrc, so I feel like it's got to be something I'm doing in my code.
>
> Thanks!
> _______________________________________________
> gstreamer-devel mailing list
> gstreamer-devel at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>
>
> _______________________________________________
> gstreamer-devel mailing list
> gstreamer-devel at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/gstreamer-devel/attachments/20190830/775e26a6/attachment.html>