Very short "phantom" timeshift buffers at the start of programs

[MrQuade]

I seem to be getting quite a few spurious changes of event on my T3 and T4 tonight.

I pressed REC on the T4 and went to select an event to save, and saw that there were about 8 entries for *MASH* showing with start times sometimes only 5 minutes apart.

I went to the T3 and checked that, and noticed that the last two events were each listed twice, and in each duplicate case, both instances of each event showed the same start time. These zero-length timeshift buffer files made it difficult to navigate back using the BACK button.

I'll keep an eye on it to see if it keeps happening, but I normally never see these sorts of spurious event changes.

[MrQuade]

MrQuade, do you have a log that covers when those zero-length (or very short) timeshift recordings were made?

I'll have a look when I am home, but I don't think I was fully set up for logging at that stage.

[prl]

MrQuade, do you have a log that covers when those zero-length (or very short) timeshift recordings were made?

I'll have a look when I am home, but I don't think I was fully set up for logging at that stage.

A pity, but if you see it again when you do have logging on, I'd be happy to have a look.

[MrQuade]

A pity, but if you see it again when you do have logging on, I'd be happy to have a look.

It seems to be happening fairly reliably.

I started the T4 up and left it for the afternoon and ended up with this just now:

Log is attached.


I'll try some other channels and see if it is happening on all of them, or if One is just having problems.

[prl]

Thanks.

[prl]

A pity, but if you see it again when you do have logging on, I'd be happy to have a look.

It seems to be happening fairly reliably.

I started the T4 up and left it for the afternoon and ended up with this just now:

[image]

Log is attached.


I'll try some other channels and see if it is happening on all of them, or if One is just having problems.

Did you take the "grab" of the log at about the same time as the screenshot? I'm trying to work out when the screenshot was taken relative to the times in the log (which are time-since-boot). I think I have a match between the timeshift start times in the log (which are relative to boot time) with the timeshift start times in the screenshot, and that puts the log "grab" at about the same time as the screenshot.

I've been running on ONE Canberra for most of the day and I'm not seeing any really short timeshift buffers. It's odd that IanSav has seen this (in a different context) in Melbourne, and you in Perth, but I don't seem to be able to find it here in Canberra.

IIRC, you're using IceTV. What is your setting for MENU>Setup>TV>EPG Settings>Show EIT now/next in infobar? I have it off, and perhaps that's a significant difference.

The fix may be similar to the fix for the start markers being moved after they've first been set.

[MrQuade]

Did you take the "grab" of the log at about the same time as the screenshot? I'm trying to work out when the screenshot was taken relative to the times in the log (which are time-since-boot). I think I have a match between the timeshift start times in the log (which are relative to boot time) with the timeshift start times in the screenshot, and that puts the log "grab" at about the same time as the screenshot.

Yep screenshot and the log were taken more or less simultaneously about 5 minutes before I posted the message.

IIRC, you're using IceTV. What is your setting for MENU>Setup>TV>EPG Settings>Show EIT now/next in infobar? I have it off, and perhaps that's a significant difference.

Yes, running Ice, and I enable EIT in the Infobar so I get the fine-grained start times.

[prl]

...

IIRC, you're using IceTV. What is your setting for MENU>Setup>TV>EPG Settings>Show EIT now/next in infobar? I have it off, and perhaps that's a significant difference.

Yes, running Ice, and I enable EIT in the Infobar so I get the fine-grained start times.

OK. I've changed my setting to on, and I'll leave it running on One Canberra, perhaps overnight, The setting affects both the times that are displayed in the infobar and dividing the stream into recordings of events in the timeshift buffer.

[MrQuade]

I didn't have a lot of luck reproducing this on ABC fit a few hours after that, but I have definitely seen it on channel 9 services too.

[prl]

I didn't have a lot of luck reproducing this on ABC fit a few hours after that, but I have definitely seen it on channel 9 services too.

I can definitely reproduce it on ONE Canberra now that I'm using the EIT for Now/Next. This one even shows the title changing on the fly:

That's also a bit of a hint as to the cause: "Attenborough's - The Life..." is from the EIT Now/Next, while "David Attenborough's - The Life..." is from the IceTV EPG.

[Grumpy_Geoff]

I didn't have a lot of luck reproducing this on ABC fit a few hours after that, but I have definitely seen it on channel 9 services too.

I can definitely reproduce it on ONE Canberra now that I'm using the EIT for Now/Next. This one even shows the title changing on the fly:

Screen Shot 2017-07-15 at 18.12.00.png



That's also a bit of a hint as to the cause: "Attenborough's - The Life..." is from the EIT Now/Next, while "David Attenborough's - The Life..." is from the IceTV EPG.

Yep, that's what I've seen in the past if I used the EIT for Now/Next and there was a slight transmission glitch, as shown here =-viewtopic.php?f=50&t=10261&start=40#p139748

[Grumpy_Geoff]

A transmission glitch can also cause it.

I enabled EIT Now/Next and tuned to ONE. I observed the Now/Next transition from MacGyver to Attenborough, then I saw a second Now/Next for the same program pop up about 4 mins after the first.
Having a look now, I can see a few more must've popped up
I'm assuming the last two Attenborough entries are from a transmission glitch, as IceTV has the program title "David Attenborough's - The Life Of Mammals", whereas FTA has "Attenborough's The Life Of Mammals".
"Undercover Boss" seems to have had a new transition after only 4 seconds.

Code: Select all

{555}< 13650.985> [SCREENNAME]  InfoBar
{555}< 13650.987> [eDVBServicePlay] timeshift
{555}< 13650.999> [eDVBServicePlay] pts of eit change: 2de48a60, fixup_pts: e3bd5e0, first_pts: 1fa8b480
{555}< 13651.004> [eDVBServicePlay] no cue sheet
...
{555}< 13885.803> [SCREENNAME]  InfoBar
{555}< 13885.804> [eDVBServicePlay] timeshift
{555}< 13885.815> [eDVBServicePlay] pts of eit change: 2f273580, fixup_pts: 14094c0, first_pts: 2de6a0c0
{555}< 13885.819> [eDVBServicePlay] no cue sheet
...
{555}< 16157.962> [SCREENNAME]  InfoBar
{555}< 16157.964> [eDVBServicePlay] timeshift
{555}< 16157.975> [eDVBServicePlay] pts of eit change: 3b55e5e0, fixup_pts: c2d42c0, first_pts: 2f28a320
{555}< 16157.980> [eDVBServicePlay] no cue sheet
{555}< 16157.992> [eDVBServicePlay] pts of eit change: 3b55e5e0, fixup_pts: c2d42c0, first_pts: 2f28a320
{555}< 16157.996> [eDVBServicePlay] no cue sheet
...
{555}< 17228.881> [eDVBServicePlay] timeshift
{555}< 17228.882> [eDVBServicePlay] pts of eit change: 41157630, fixup_pts: 5b9d9d0, first_pts: 3b5b9c60
{555}< 17228.887> [eDVBServicePlay] no cue sheet
...
{555}< 17232.630> [eDVBServicePlay] timeshift
{555}< 17232.631> [eDVBServicePlay] pts of eit change: 411abc30, fixup_pts: 278d0, first_pts: 41184360
{555}< 17232.635> [eDVBServicePlay] no cue sheet

[prl]

I think I have a handle on where the problem is coming from, it's now trying to work out how to do it right.

[MrQuade]

Sweet

[MrQuade]

I appreciate the reason for moving the discussion to its own thread, but at the same time, I have never noticed this issue prior to the 17.5 beta. I had thought that it ma have been the result of some if the other event change detection optimisations. Happy to be corrected though.

[prl]

I appreciate the reason for moving the discussion to its own thread, but at the same time, I have never noticed this issue prior to the 17.5 beta. I had thought that it ma have been the result of some if the other event change detection optimisations. Happy to be corrected though.

Grumpy_Geoff's link to the same problem in his earlier post goes back more than a year to May 2016.

The changes I made to fix up the start/end marks in recordings are in a different part of the code and shouldn't impact on this issue.

The problem comes down to differences in event ids and start times between the now EIT data and the EPG data for the same program.

I've also noticed that the different event ids between the now EIT data and the IceTV EPG mean that the .eit files for timeshift recordings aren't created if you use IceTV and enable Show EIT now/next in infobar.