[Bug 581011] Move DVB plugins to -good

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=581011
  GStreamer | gst-plugins-good | 0.10.x

Fabrizio Milo <mistobaan> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mistobaan at gmail.com

--- Comment #4 from Fabrizio Milo <mistobaan at gmail.com> 2011-06-10 01:33:52 UTC ---
> dvbsrc should have a device= property, like v4l2src.
how would be different from the adapter and frontend properties ?
one adapter(ie. device) could potentially have multiple frontend.

> tuning needs to be asynchronous. 
Agreed.

> Holding a lock and not returning from g_object_set()
> until tuning is finished is not ok.
Agreed. 

> - What is the most appropriate thread to do tuning?  In PLAYING,
>  probably streaming thread.  In READY, probably special thread.

> Restore timeout property from 5616efb0.
Yes there are many timeouts inside the logic, one for the poll, one for
acquiring the signal, one for retrying. and are all "timeout" but
I think the main timeout sense is how long should wait before giving
up in tuning. right ?

> polarity property should be an enum (or boolean).  Or autoprobed.
> Needs documentation of dvb-adapter structure.
> Needs documentation of dvb-frontend-stats structure.
ok.

> Perhaps should have a URI handler
Are you talking about : dvb_base_bin_uri_set_uri inside dvbbasebin.c ?


> Move gst_dvbsrc_plugin_init() somewhere else.
why ? 

> In read_dvb_device(), why are there two timeout paths?
to differentiate from a single poll timeout and a total timeout
(i.e when the signal drops). Requires a bit more of logic
and I think needs to be considered with bug 641204

> In read_dvb_device(), it would be better to configure the device not
> to wake up until there are N packets to read, and/or some delay has
> passed.  This avoids unnecessary read()/poll() syscalls.  Is this
> possible?
I am not sure. But I agree on the strategy.

> Buffers should be timestamped by the hardware.
Not sure how this can be achieved. The mpegts packet contain timestamps 
usually but are not handled at this level.

> What is the purpose of gst_dvbsrc_start_stop_filters()?  Please document.
I guess was for be able to add or remove at runtime PIDS to be monitored by the
hardware.

> gst_dvbsrc_frontend_status() has a usleep().  This should be removed.
Agreed

> Move diseqc to separate source file?
Yes seems does not belong to the dvbsrc class too much, they are a separate set
of commands.
Should we remove also those usleep from the logic ?

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.