Anomalies using Sage.jar v9.0.0.3 in SageTV7

[Monedeath]

Originally Posted by sacrament055

Not sure if this is related but I thought I'd mention it here. I have 5 ATSC tuners and I have Merit set to prefer my higher quality tuners and that has been working as far as I know until I moved to the new sage 9 jar.

Not quite sure if the Merit is not opposite (lower number is preferred) or if it's random but I keep seeing it pick a tuner that has Merit of 0 instead of a tuner that has Merit of 7 when there's nothing else recording.

I think I'very only seen Sage get weird with tuner merit(seemingly ignoring it) when it comes to use of Sage as a live TV viewer, scheduled recordings seem to still adhere to the logic properly. Are you seeing the same thing or is it ignoring merit on scheduled records as well?

(Configured with three tiers in my case, the Hauppauge 2250 has highest merit, followed by the two HD Homeruns before it is supposed to try for the Ceton tuner which has the lowest tuner merit in my setup... but where the local channels overlap, tuning into live TV with no active recordings going, we could just as easily end up on the Ceton or Homerun rather than the 2250. Scheduled records behave mostly as expected when it comes to tuner selection, scheduler randomly becoming convinced no other tuners are available when several are not withstanding.)

[sacrament055]

I'll have to go back and check my recordings but I think it's picking the same tuner by default for recording. Oddly I removed one of my preferred tuners and re-added it and now Sage is picking it first. I'm thinking I might try removing my tuners from the sage.properties file and re-add them and reconfigure merit to see if that resolves it.

It hasn't been that big of a deal so far because none of my tuners are terrible but 2 of them are slightly better at tuning in weak signals or multipath issues.

[Narflex]

Originally Posted by wnjj

I think I figured out the Actor/Character repeating issue.

There were new fields added to the "Person" table for wiz.bin version 0x4F and above. My 7.1.9 version was using 0x4B and V9 is 0x55.

    if (ver >= 0x4F)
    {
      Wizard wiz = Wizard.getInstance();
      extID = in.readInt();
      .
      .
      .

The extID field is used to match actors to their roles in SeriesInfo.java:

  public String getCharacterForActor(Person actor)
  {
    if (actor == null) return "";
    for (int i = 0; i < people.length; i++)
    {
      // Resolve aliases as well so they match
      if (Math.abs(people[i].extID) == Math.abs(actor.extID))
        return characters[i];
    }
    return "";
  }

The issue is when the V7 db upgrades to V9, all of the actors have an extID of 0. So when the match takes place, the first actor in the DB matches the ID of the actor in the show. So what's needed is if the actor extID is 0, fall back to the other string search "guessCharacterForActor" instead:

  public String getCharacterForActor(Person actor)
  {
    if (actor == null) return "";

    if (actor.extID == 0) return guessCharacterForActor(actor.name);

    for (int i = 0; i < people.length; i++)
    {
      // Resolve aliases as well so they match
      if (Math.abs(people[i].extID) == Math.abs(actor.extID))
        return characters[i];
    }
    return "";
  }

Nice catch. I'm assuming you'll submit a commit to correct this...if you'd rather I just do it, let me know..but I don't want to just steal your credit for fixing the issue without your permission.

[wnjj]

Originally Posted by Narflex

Nice catch. I'm assuming you'll submit a commit to correct this...if you'd rather I just do it, let me know..but I don't want to just steal your credit for fixing the issue without your permission.

I have no problem with you just fixing it. Mainly because there are a few other person functions in there relying on extID and I didn't look at the impact of those. I also haven't completely gotten my head around the id versus extid concept. I know id is an internally generated number inside Wizard. I assume extID comes from the EPG and is meant to be more universal but couldn't find exactly where that gets populated. When a SeriesInfo gets built, it looks like the people[] array get populated with Wizard id's by looking up names. So should the character lookup function fall back to matching on people[i].id==actor.id rather than using a text match?

Also, what do the negative id's mean? Are negative id's used to look in indexes and positive ones used directly as id's?

[Monedeath]

I guess we'll have to see if that fully fixes things, once someone compiles a new jar file. But I would point out that the credits issue happens on new recordings made by 9.0.0.3 as well as the older recordings from previous builds of Sage. The screenshot provided by me earlier was a 9.0.0.3 recording. (" Issue is present on new and old recordings alike." In my initial report)

Good work finding where that lives in the code regardless. Although there probably are a few more moles lurking about to get whacked in this round of whack-a-mole.

[wnjj]

Originally Posted by Monedeath

Good work finding where that lives in the code regardless.

Thanks for that.

I wonder if the stuff that populates extID somehow got removed during the open source port. If that's the case we'd want to switch all extID back to just id.

The way I found this is by looking through the java code a bit after I ran my latest creation. I build a standalone C++ program that can read wiz.bin. Right now it doesn't do anything useful with the data but I can slap debug printf's in to trace things like this. Iteration time is mere seconds using a standalone program. Eventually I'd like to send the data off to pure txt or even SQL for import into a completely relational DB. Then I'd finish out with a wiz.bin writer. This may eventually allow for "fixing" corrupted wiz.bin files by hand.

[Narflex]

ExtID for people is not something that'll get populated in the open source build. That was using extra data we had on Google Fiber that's not available on the SageTV EPG server. It's possible one day that Schedules Direct will have that data...which is why I left that code in there. The value should be zero for extID for all Person objects for everyone except me (I have a special plugin that uses the Google Fiber EPG server...and no, I can't share it)

The negative values for extID that you're seeing were for 'aliases'. This is for when there's multiple names associated with the same person and the primary name had the positive ID; and the negative ones (with the same absolute value for ID) were the alias names. We used this to link them all together in a way that was fast to resolve.

I'll also go ahead and make that code fix now too.

[Narflex]

OK, change is in now.

[Slugger]

Originally Posted by Narflex

ExtID for people is not something that'll get populated in the open source build. That was using extra data we had on Google Fiber that's not available on the SageTV EPG server. It's possible one day that Schedules Direct will have that data...which is why I left that code in there.

Schedules Direct has support for all that extra data. My client doesn't make use of it, but that's only because I had no need for it at the time. The data is available if/when the client is extended to pull it in. (The data is already in the sdjson.epg cache file sitting on your Sage box if you're using the plugin, the client just needs to be updated to access it and do "whatever" needs to be done to put it into wiz.bin. I assume there are Sage apis around that area that I just haven't read up on yet).

[Denali]

Thanks wnjj and Narflex for solving the Actor anomaly. Time to give version 9 another try when a new sage.jar is built.

[wnjj]

Originally Posted by Narflex

OK, change is in now.

Do hasActorInCharacterImage and getActorInCharacterImageURL need updated for the same issue or is CharacterImage data not applicable for the OS version either? It looks like they aren't used internally but are exposed in the API.

[Narflex]

Originally Posted by wnjj

Do hasActorInCharacterImage and getActorInCharacterImageURL need updated for the same issue or is CharacterImage data not applicable for the OS version either? It looks like they aren't used internally but are exposed in the API.

The current EPG downloaders won't have the information needed to resolve those into URLs. It might be possible to update the SD plugin to do that...but I'm not aware if they have that data or not.

[Slugger]

SD has all that data.... the SD plugin for Sage does not use it. The client interface to the data would require enhancements to be able to access that data. I have no plans to do that anytime soon (i.e. definitely not before 2016, my first Sage OSS project is lined up and is going to take some time and that's all I'm going to have time for into 2016), but all the code is on github for anyone wanting to dive in. But I don't think pulling the data into wiz.bin is enough, I assume no STVs make use of this data either so keep that in mind too.

[Opus4]

Originally Posted by Slugger

SD has all that data.... the SD plugin for Sage does not use it. The client interface to the data would require enhancements to be able to access that data. I have no plans to do that anytime soon (i.e. definitely not before 2016, my first Sage OSS project is lined up and is going to take some time and that's all I'm going to have time for into 2016), but all the code is on github for anyone wanting to dive in. But I don't think pulling the data into wiz.bin is enough, I assume no STVs make use of this data either so keep that in mind too.

Actually, the STV I'm almost ready to update in GitHub has the ability to display images for people. If you aren't getting back to working on the SD EPG data until 2016, then I suppose that means there's a chance I might get to it before you, but I have another feature I would like to add first too.

Andy

[egeller]

Originally Posted by Denali

2. Some EPG shows fail to show Future Airings.

I am still seeing this for "MLB Baseball". I can see the past, but not future, entries. Is there something I missed or a work-around?

[robogeek]

Originally Posted by egeller

I am still seeing this for "MLB Baseball". I can see the past, but not future, entries. Is there something I missed or a work-around?

I see this too with some shows, running SageTV v9. It seems like it may be related to the number airings (or watched airings) a show has had. My wiz.bin is about 10 years old, so it has tracked many airings. When the blank Future Airings list occurs, the Past Airings list is always capped at exactly 1000 items. Seems like there is some sort of limit being imposed under certain circumstances and when Past Airings=1000, Future Airings fail to show. I see it happen with shows like Jeopardy, Wheel of Fortune, and Family Feud (all set as Favorites). But a local news broadcast I have set as a favorite has over 3200 past airings in the Past Airings list, and some other long-running TV shows seem to be alright. Though some of the long-running TV shows (like daytime soap operas that have had more than 1000 airings) seem to be having their Past Airings item list pruned down to under 1000 so the blank Future Airings doesn't occur with those shows.

[Narflex]

Originally Posted by robogeek

I see this too with some shows, running SageTV v9. It seems like it may be related to the number airings (or watched airings) a show has had. My wiz.bin is about 10 years old, so it has tracked many airings. When the blank Future Airings list occurs, the Past Airings list is always capped at exactly 1000 items. Seems like there is some sort of limit being imposed under certain circumstances and when Past Airings=1000, Future Airings fail to show. I see it happen with shows like Jeopardy, Wheel of Fortune, and Family Feud (all set as Favorites). But a local news broadcast I have set as a favorite has over 3200 past airings in the Past Airings list, and some other long-running TV shows seem to be alright. Though some of the long-running TV shows (like daytime soap operas that have had more than 1000 airings) seem to be having their Past Airings item list pruned down to under 1000 so the blank Future Airings doesn't occur with those shows.

This makes perfect sense as the search API is used to find these items. If you go into Detailed Setup->Customize and change the "Maximum Results per Search" value (I set mine to 10000) then it'll fix this. If you're using an extender though, you'll need to manually set it in the Sage.properties file on the server though which is: wizard/max_search_results

[Denali]

Thanks Narflex for this info. It was last reason for not upgrading to version 9. In SageTV7.xml a minor problem with multiple blank lines between actors in "View Series Information" corrected by changing

Action Name="" \n \n \n""
to
Action Name="" \n""