Dreamboard

FoxyRabbit · 20. August 2017

Hi,

Is this a known problem for everybody, or is it pr satellite problem. Nordic 0.8 and 4.8 will alway show <unknown> for TTX subtitles. Have not tested other satellites yet...

FoxyRabbit · 21. August 2017

Someone told me that OpenPli images have a solution for this, and we have checked the same channel with both Dream Property image and Openplis's 6.0 image. DP image shows <unknown> for 10 SVT1 subtitle ttx pages. OpenPli shows teletext page numbers for all 10 TTX pages.

I have gone through som coding and found that enigma2.py for openpli have used attribute 'subtitle" from iPlayableServicePtr to catch page numbers. This attribute is not available in enigma2.py from DP image.

dre · 21. August 2017

Since when is OpenPli using the oe2.5 as basis? Please do not compare apples with pears. Especially as there were a lot of changes in that area with oe2.5

FoxyRabbit · 21. August 2017

The OpenPli may be a damn bad comparsion, and I also asumed they used oe2.5 for their latest images. My bad!

So basically, what you are telling me is that it is not possible to do the same thing in oe2.5? Or that it takes more time to develope a solution for it? Are you working on it? Was you aware of this?

We are just wanting to help you guys, you know.

m0rphU · 21. August 2017

The fact that something is missing in enigma2.py indicates a missing feature (or a bug) in the enigma2 C++ core. That part is closed source in DreamOs and can only be enhanced by the Dream developers. As @dre is an independent developer (and just a regular user like you and me), he does not have any access to the core.

Another side fact about your bug report: The Dream developers usually read these kind of reports and put them on their own, internal bug tracker. We don't know which priority these bugs have there and usually the first response you will receive, will be a remark in the changelog of the update that fixes your problem

Last, but not least: My personal view on this is, that a fix might take some time. Germans are the main target group for Dream and they don't use subtitles at all. As a comparison: The German public tv stations have 1 TTX-subtitle and the same one as DVB subtitle... Most of the other channels have no subtitles. Therefore 10 different subtitles are hard to imagine for us

dre · 21. August 2017

I never use subs but I thought that the page number was displayed on SRF. I'll have to check it once I'm back home

FoxyRabbit · 21. August 2017

Thanks guys! Lets hope they fix this once upon a time. An lets hope that such a small fix can make others than germans to use dreamboxes. Up in north we have tons of ttx subtiles on lots of channels. Pretty anoying to have to test every ttx subtitle to find what you want.

I think this was no issue on oe2.0 and earlier versions, but may have come with the oe2.2 images. It has been an issue for a while for sure.

Meanwhile I just discovered that two channels on 4.8 actually shows language (countries) for TTX subtitles. So it might only be the subtitles with TTX page numbers that are missing. Have not seen any channels with that yet.

As far as my knowledge in python gets, I see that the only thing returned for TTX is when data have language names in it. Else it will return "und". Its for these returned strings we need to do something extra to get the page numbers. Such code is missing as I can see.

dhwz · 21. August 2017

Thats the real issue, the new audioselection screen only shows a column with the ISO language codes. This is completely different to OpenPLi. The root cause is that your providers don't provide proper language settings for each subtitile page. I think its also not as easy to fix (just displaying the teletext page number) as this may break the Trackautoselect plugin then.

mr_vica · 21. August 2017

habe momentan PLi am Laufen, der einzige Sender auf Astra wo man sich das anschauen kann ist TV5MONDE DVB-S 11538V, 22000, 5/6, QPSK, 2 TTX und 7x DVB Untertitel

dhwz · 21. August 2017

Das ist ein schlechtes Beispiel weil der Sender macht alles richtig

mr_vica · 21. August 2017

Nordic 0.8 und 4.8 kann ich nicht anpeilen
@FoxyRabbit, can you make some snapshots of the subtitle selection

FoxyRabbit · 21. August 2017

Since older software could catch these numbers, it should be possible to implement this in newer software also. I can understand if its not "developed" yet for oe2.5, but it should be possible to do the same thing as long as older software can do this from todays satellite providers. As OpenPli does.

Cant really see why getting the numbers out should mess with trackautoselect, as I asume it also will do that with only <unknown> returned. And as long as it returns numbers, you could filter them out by leaving numbers out of the logic. Cause you will most likely never know what subtitle laguagea a number returns anyway. For a tuner owner it might give a big logic, cause these numbers often is the same for each providers channels. Or are remembered by the user for spesific channels.

All numbers in first column that mr_vice shows are the numbers that are not returned in oe2.5.

So if its possible, it should be fixed. And in my and many others opinion it is an important issue. Maby not for germans, but for plenty other countries that wants to use dreambox tuners.

FoxyRabbit · 21. August 2017

Here is some pictures from Peter Pan 2017 - stealth (beta) vs. OpenPli 6.0 (not a dreambox, but E2 anyway)

<ukjent> (norwegian) is the same as <unknown> in english by the way.

Same channel SVT1

FoxyRabbit · 22. August 2017

When I choose one <unknown> the subtitle comes around. So I guess it has to return the right number for teletext to use...

mcs1 · 22. August 2017

Hi,
There seems to be areal issue with the subtitles and the explanation is given above. I think this may well be the same problem I reported with the timeshift a few days ago, timeshift gives a green screen and some integer exception with subtitles on.
Cheers
Mcs

Ghost · 22. August 2017

Hi,

can you try the following?

Python

diff --git a/usr/lib/enigma2/python/Screens/AudioSelection.py b/usr/lib/enigma2/python/Screens/AudioSelection.py
index 149d936..036c65b 100644
--- a/usr/lib/enigma2/python/Screens/AudioSelection.py
+++ b/usr/lib/enigma2/python/Screens/AudioSelection.py
@@ -256,6 +256,9 @@ class AudioSelection(Screen, ConfigListScreen):
                                        flags.append(_("Saved"))
                                s_flags = (", ").join(flags)
+                               if s_codec == "TTX":
+                                       s_codec += " %x%x" %(trackinfo.getMagazineNumber(), trackinfo.getPageNumber())
+
                                stream = (SelectionTrackinfoEntry(idx, trackinfo), s_number, s_language, s_codec, s_flags, selected and selectionpng or None)
                                streams.append(stream)

Alles anzeigen

I think this is okay.

cu

FoxyRabbit · 22. August 2017

Thanks @Ghost!

This returns a better result than OpenPli in my opinion. By adding TTX numbers to codec text column was a smart one. Cause we may still dont know which language we get for each teletext page. BUT, there may still be a small issue with some page numbers (getPageNumber) returning hex numbers. Like TTX 7fe or 1a0 or 1fe as exampels. SVT1 channel with 10 TTX pages is the only channel I have found with some hex numbers of the 10 numbers. It is always the last 2-3 numbers or so.

Attachment: picture of channel SVT1 with unknown language, and Viasat Film Action with known language.

Ghost · 22. August 2017

Because of the missing languages. The problem is that these subtitle tracks are not included in the PMT.

The TTX subtitle tracks without language are automatically recognized as subtitles by the teletext decoder.

But unfortunately without any information about the language.

I have no knowledge about a way to detect the language.

If you want, you can send a message to the provider, and ask him to put the subtitles into the PMT of the service with corresponding language codes.

cu

FoxyRabbit · 22. August 2017

This is just perfect until the provider implements language strings for each teletext number. It was the numbers we wanted. By adding the numbers to codec text, everybody will know that the language for the number still is <unknown>. Perfect!

Over time users wil memorize numbers with different languages, so starting with showing the numbers are just perfect. In most cases the same number will be used for same language anyway. Our image users complained about this, and now they will be satified I think. But yes, we will try to contact the provider for missing language strings.

Will you patch this into git some day? I think more out there than me will be more than satisfied with this "fix". One less file for me to play with in my repository would be nice too.

Edit: By the way, if I leave the subtitle setup screen on TV for more than 5 minutes, the HEX numbers seems to correct itself!