[mythtv-users] Request for help making live TV useable

[Mark Lord]

On 11-04-17 08:55 AM, Mark Lord wrote:
>
> I have the same devices as the OP, plus others, and I agree 99% with his assessment.
> 0.24-fixes is far, far worse for LiveTV than 0.22, 0.21, 0.20, 0.19, and 0.18 were
> with this setup.
> 
> I don't have any trouble with liveTV and DVB tuners, nor with the HDHomeRun,
> but anything to do with NTSC analog has issues.  The ConvertX doesn't work at all
> in LiveTV (fine for recording, though), none of my "soft" encoders work
> beyond the first channel shown (if that), and even the PVR-250 results in lockups
> and FE crashes if I change channels too quickly (more than once every 10 secs or
> so).
> 
> Dunno why, and don't care enough to spend hours figuring it out and fixing it,
> but it's there and not just an isolated incident.  (I do spend hours finding
> and fixing other stuff, though, just don't have enough of those to worry much
> about LiveTV).
> 
> I do see that Mark Kendall has a new LiveTV fix in 0.24-fixes today though,
> so I'm downloading/rebuilding with that to see if it has any effect here.


I can still get LiveTV (NTSC OTA) to lock up with about the same
degree of reliability as before.  So another small step perhaps,
but somethings still not quite right in there.

However.. I also noticed a bunch of decoder fixes in trunk,
dealing with calculating the number of buffers needed,
and fixing bugs where buffers were being recycled while
still in use and so on.

So I backported those as well to 0.24-fixes.  This seems to have made
a noticeable improvement in my PVR-250 LiveTV channel changes.
I can now often change channels 8-10 times without locking up,
whereas normally only (exactly) three channel changes were needed before.

Those patches are definitely a keeper on the machines here!
And they have the side benefit of no longer requiring that I
hardcode vdpaubuffersize=nn on various machines.  Woo-hoo!  :)

Now.. something weird since I upgraded from 0.22 to 0.24:
When I view LiveTV (for analog OTA NTSC), there are always *two*
files created for each "recording", one of which is a zero-length file,
and the other (created second) has the actual recording.

Is this "normal" ?

Thanks