[mythtv] [mythtv-commits] [MythTV/mythtv] cf8015: No discontinuity for first TS packet of PID

[Klaas de Waal]

On Sat, 29 Aug 2020 at 21:43, David Engel <david at istwok.net> wrote:

> On Sat, Aug 29, 2020 at 07:36:37AM -0700, kmdewaal wrote:
> >   Branch: refs/heads/master
> >   Home:   https://github.com/MythTV/mythtv
> >   Commit: cf8015eaf4a2fde6ece423e7ade99c86f7f9d076
> >
> https://github.com/MythTV/mythtv/commit/cf8015eaf4a2fde6ece423e7ade99c86f7f9d076
> >   Author: Klaas de Waal <kdewaal at mythtv.org>
> >   Date:   2020-08-29 (Sat, 29 Aug 2020)
> >
> >   Changed paths:
> >     M mythtv/libs/libmythtv/recorders/dtvrecorder.cpp
> >     M mythtv/libs/libmythtv/recorders/dtvrecorder.h
> >
> >   Log Message:
> >   -----------
> >   No discontinuity for first TS packet of PID
> >
> > Do not count the discontinuity from 0 to the continuity counter value of
> the first transport stream
> > packet of a PID as an error and consequently do also not give a
> "discontinuity detected" error message.
> > A perfect recording will now report a continuity_error_count of 0
> instead of approximately the number of PIDs in the recording.
>
> By sheer coincidence, I've been debugging an issue with my setup and
> have been watching for continuity errors.  I was lamenting the fact
> that there were often false positives reported every time a recording
> starts when I saw you committed this change.  I see one problem with
> the change, though.  1/16th of the time, the first counter seen will
> actually be 0, which is the inital value in m_continuityCounter.  When
> that happens, reporting of the first, real discontinuity will be
> suppressed.
>
> I believe the attached patch fixes that issue by initializing
> m_continuityCounter to an impossible value.  That will always cause an
> initial continuity that can be safely ignored.  All continuity errors
> after that will be real and can be reported.
>
>
> Hi David,

Your analysis that 1/16th of the time the first real error is suppressed is
of course correct.
I do appreciate your patch but it does need initialisation of all elements
of m_continuityCounter to 0xff, not only the first element.
This can be done by calling m_continuityCounter.fill(0xff) in the
constructor, I cannot find a way to do this in the header file.
I am testing this now and will commit it later.

Thanks,
Klaas.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mythtv.org/pipermail/mythtv-dev/attachments/20200829/e1d88e50/attachment.htm>