finish GstDiscoverer impl and add exampl

examples/decodebin/main.go

@@ -1,3 +1,33 @@
+// This example demonstrates the use of the decodebin element
+// The decodebin element tries to automatically detect the incoming
+// format and to autoplug the appropriate demuxers / decoders to handle it.
+// and decode it to raw audio, video or subtitles.
+// Before the pipeline hasn't been prerolled, the decodebin can't possibly know what
+// format it gets as its input. So at first, the pipeline looks like this:
+
+// {filesrc} - {decodebin}
+
+// As soon as the decodebin has detected the stream format, it will try to decode every
+// contained stream to its raw format.
+// The application connects a signal-handler to decodebin's pad-added signal, which tells us
+// whenever the decodebin provided us with another contained (raw) stream from the input file.
+
+// This application supports audio and video streams. Video streams are
+// displayed using an autovideosink, and audiostreams are played back using autoaudiosink.
+// So for a file that contains one audio and one video stream,
+// the pipeline looks like the following:
+
+//                        /-[audio]-{audioconvert}-{audioresample}-{autoaudiosink}
+// {filesrc}-{decodebin}-|
+//                        \-[video]-{viceoconvert}-{videoscale}-{autovideosink}
+
+// Both auto-sinks at the end automatically select the best available (actual) sink. Since the
+// selection of available actual sinks is platform specific
+// (like using pulseaudio for audio output on linux, e.g.),
+// we need to add the audioconvert and audioresample elements before handing the stream to the
+// autoaudiosink, because we need to make sure, that the stream is always supported by the actual sink.
+// Especially Windows APIs tend to be quite picky about samplerate and sample-format.
+// The same applies to videostreams.
 package main
 
 import (