skywalker-1 tuner scan all dish channel get unknown name (1 Viewer)

[mm1352000]

I'm happy to have a go at fixing it (although I'll probably need to ask a few questions along the way, since I know very little about the internals of TsWriter/the scanning process).

I'm happy to advise.
If my guess is correct it should be quite straightforward. In PatParser.cpp in OnSdtReceived, change:

  // check if we already set the sdt for this channel
   if (info->SdtReceived) return;

...to...

  // check if we already set the sdt for this channel
   if (info->SdtReceived || info->TransportId != sdtInfo.TransportId) return;

In the meantime, would it be OK to attach a 'rolled-back' version of MP 1.5 TsWriter here for the benefit of the sat users with the scanning problems ?

Sure, this is open source - you can do what you like! No need to ask for my approval.
Having said that, personally I think it would be better to take a look at the Dish dump to figure out why the EPG is not linked up with the channels correctly. The above modification may well remove the requirement for modifying the TSID. I'm just not sure about the ONID thing.

[edit: ...but having said that, the NID is set from the SDT so if you select the correct SDT record then the NID should be set correctly. I just wonder if Dish has multiple channels with the same SID + TSID combination and different NID values. That would be strange. Hmmm... maybe the above modification will solve all the problems... ]

 

[Owlsroost]

Having said that, personally I think it would be better to take a look at the Dish dump to figure out why the EPG is not linked up with the channels correctly

What tool do you use to look at the ts dump files ?

 

[mm1352000]

Okay, I can confirm that tweak gives me full channel details - names and correct LCNs - for NZ Freeview HD, and it also gives me channel names for the Dish channels when I scan the provided dump with a high enough SDT/VCT timeout. I note that the provider names for some of the Dish channels are a bit strange, and the channels also don't get LCNs.

On further investigation of the Dish stream, I think the SDT actual is being carried in the adaptation field in the SDT PID stream. No idea why they'd choose to do that. If we were able to parse the SDT actual out of the adaptation field it would probably result in faster scanning... but I'm not going to attempt that change in the foreseeable future. LCNs could probably be figured out with a bit of time. The other thing that would be good is to add the Dish custom service types into TV library. All potential improvements for the future...

 

[Owlsroost]

OK, I'll try it on UK Freeview.

 

[mm1352000]

What tool do you use to look at the ts dump files ?

Depends what I'm doing. Mainly 3 tools:

1. Notepad++ with the HEX-Editor plugin - for deep packet inspection and searching for specific content.
2. TransEdit (DVBViewer scanning/analysis tool) - for general checks on SI, EPG, descriptors etc.
3. Silicondust HDHomeRun simulator - for running TS dumps through TV Server.

 

[Owlsroost]

OK, I'll try it on UK Freeview.

Just done a quick test - all seems to be working OK with the mod - all channels have LCN's, and the EPG grabbing is OK. Thanks

I'll do some longer testing later today.

 

[Owlsroost]

Further testing didn't show up any problems (except that once - out of probably 10 or more scans - I got a few '10000' channels, but nothing in the logs to indicate these were from 'otherMux' data, so I suspect that was due to some other glitch as I can't reproduce it).

So attached are TsWriter versions with mm's mod included for MP 1.4.0 and MP 1.5.0 - please test and report back please....

(Please note - I've tested the MP 1.5.0 version, but not the 1.4.0 version)

 

[mm1352000]

Further testing didn't show up any problems (except that once - out of probably 10 or more scans - I got a few '10000' channels, but nothing in the logs to indicate these were from 'otherMux' data, so I suspect that was due to some other glitch as I can't reproduce it).

Probably NIT not received within the scan time. NIT is usually the slowest cycle time of all the tables we grab from, so its the easiest to miss.

 

[Owlsroost]

@liewfr and @wderbi - it would be great if you could test the modified TsWriter.ax please

 

[wderbi]

@liewfr and @wderbi - it would be great if you could test the modified TsWriter.ax please

Sure....I am just now seeing this. I can do this within the next day or so. Just let me know exactly what I have to do after I unzip to use the new file you posted. Also, I could upload a new TSdump if you'd like.

Just let me know...